HyBR: A Hybrid Bio-inspired Bee swarm Routing protocol for safety applications in Vehicular Ad hoc NETworks (VANETs)

Journal of Systems Architecture xxx (2013) xxx–xxx

Contents lists available at SciVerse ScienceDirect

Journal of Systems Architecture journal homepage: www.elsevier.com/locate/sysarc

HyBR: A Hybrid Bio-inspired Bee swarm Routing protocol for safety applications in Vehicular Ad hoc NETworks (VANETs) Salim Bitam a,⇑, Abdelhamid Mellouk b, Sherali Zeadally c a

LESIA Laboratory, Department of Computer Science, University of Biskra, P.O. Box 145, 07000 Biskra, Algeria Networks & Telecommunications Department and LiSSi Laboratory - IUT C/V, University of Paris-Est Créteil VdM (UPEC), Vitry-sur-Seine & Creteil, France c Department of Computer Science and Information Technology, University of the District of Columbia, Washington, DC 20008, USA b

a r t i c l e

i n f o

a b s t r a c t

Article history: Available online xxxx

Increasing interests in Vehicular Ad hoc NETworks (VANETs) over the last decade have led to huge investments in technologies and research to improve road safety by providing timely and accurate information to drivers and authorities. To achieve the timely dissemination of messages, various routing protocols for VANETs have been recently proposed. We present a Hybrid Bee swarm Routing (HyBR) protocol for VANETs. HyBR is based on the continuous learning paradigm in order to take into account the dynamic environmental changes in real-time which constitute a key property of VANETs. The protocol combines the features of topology routing with those of geographic routing. HyBR is a unicast and a multipath routing protocol (aimed at both urban and rural scenarios) which guarantees road safety services by transmitting packets with minimum delays and high packet delivery. To demonstrate the effectiveness and the performance of HyBR, we conducted a performance evaluation based on several metrics such as endto-end delay, packet delivery ratio, and normalized overhead load. We obtained better performance results with HyBR in contrast to results obtained from traditional routing algorithms such as Ad hoc On-Demand Distance Vector (AODV) topology-based routing protocol and Greedy Perimeter Stateless Routing (GPSR) geography-based protocol. Ó 2013 Elsevier B.V. All rights reserved.

Keywords: Vehicular Ad hoc Network Multipath routing Bio-inspired computing Bee swarm Road safety service Performance

1. Introduction The design of new Intelligent Transportation Systems (ITSs) is playing a major role in improving road safety, traffic monitoring and comfort for passengers in order to avoid accidents and traffic congestion [34]. To achieve these goals, ITSs need to support the delivery of timely and accurate information to drivers and authorities through a reliable Vehicular Ad hoc NETwork (VANET). VANET is considered to be a special kind of Mobile Ad hoc NETwork (MANET) with specific characteristics [44]. In VANETs, nodes are vehicles which move according to a restricted mobility pattern based on many factors such as road course, encompassing traffic and traffic regulations [37]. VANET supports communications among vehicles via Inter-Vehicle Communication (IVC) and between vehicles and fixed Road Side Unit (RSU) equipment through Roadside-to-Vehicle Communication (RVC). RSUs can be deployed at critical locations such as slippery roads, service stations, dangerous intersections or places well-known for hazardous weather conditions [8]. Nevertheless, unpredictable and inconsistent relative ⇑ Corresponding author. Tel.: +1 2405435778. E-mail addresses: [email protected] (A. Mellouk), [email protected] (S. Zeadally).

(S.

Bitam),

[email protected]

node velocity may cause intermittent link breakages. Moreover, a node in VANETs can be equipped with a Global Positioning System (GPS) device to easily determine its own location. One of the most important aspects that determines the success of VANET is the reliable message routing from a source node to a destination node. Routing in VANET relies on the presence of a sufficient number of VANET nodes that constitute strong paths to allow the forwarding of messages in the network. These paths can be affected by the vehicles’ mobility and traffic density, frequent network topology changes making them unsustainable and unreliable [20]. Therefore, the design of an efficient routing protocol for VANET is considered to be a critical issue. Moreover, one of the most important requirements in the routing process is to share integrated data with road safety service in real time in order to provide the information passengers need to help them make safe decisions. Service guarantees are important in delivering messages with a maximum packet delivery ratio on one hand, and on the other hand with a minimum routing overhead and end-to-end delay which have become a challenge for most routing protocols for VANETs. Route discovery and maintenance can affect the requirements of safety applications [14]. In this work we focus on the following issue: when communication end-points are not within their respective radio transmission range, how is it possible to

1383-7621/$ - see front matter Ó 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.sysarc.2013.04.004

Please cite this article in press as: S. Bitam et al., HyBR: A Hybrid Bio-inspired Bee swarm Routing protocol for safety applications in Vehicular Ad hoc NETworks (VANETs), J. Syst. Architect. (2013), http://dx.doi.org/10.1016/j.sysarc.2013.04.004

2

S. Bitam et al. / Journal of Systems Architecture xxx (2013) xxx–xxx

establish communication between two vehicles or between a vehicle and a roadside base station which satisfies the constraints imposed by road safety applications? To address this challenge, we propose a new bio-inspired routing protocol called the Hybrid Bee swarm Routing protocol for VANET (HyBR). It is a hybrid protocol that combines geographic routing based on Global Positioning System (GPS) to establish routes, with topology-based routing which discovers paths using network topology data. HyBR is based on the continuous learning paradigm. 1.1. Contributions of this paper To decrease the routing overheads frequently incurred by traditional routing protocols, HyBR disseminates the transmitted packets in a stochastic manner [10]. In addition, HyBR uses multiple paths simultaneously between the source and the destination to send packets in order to reduce the transmission time (end-toend delay), to decrease routing overhead and to increase packet delivery ratio. Consequently, HyBR guarantees data transmissions in real time to help drivers make safe decisions and to improve road safety. To validate the effectiveness and performance of HyBR, we have implemented HyBR using the network simulator ns-2 [33]. We conducted our performance evaluation by using realistic scenarios of VANETs through ns-2. We compare the performance of our proposed HyBR approach with well-known past routing approaches such as Ad hoc On-Demand Distance Vector routing protocol (AODV) [36] and Greedy Perimeter Stateless Routing (GPSR) [23] considered as standard topology-based and geography-based routing protocols respectively. Our performance evaluation experiments are based on performance metrics such as the average endto-end delay, the packet delivery ratio, and the normalized routing overhead. The rest of this paper is organized as follows. Related work is presented in Section 2 followed by Section 3 which describes our proposed hybrid routing protocol HyBR for VANETs. In Section 4, the experimental evaluation of proposed HyBR approach is explained. Section 5 presents the performance results obtained. Finally, Section 6 concludes the paper and highlights areas of future work. 2. Related work Recently proposed routing protocols for VANETs can be broadly classified into two categories. One category is topology-based and uses network topology data to connect vehicles, and the other category of protocols called Geography-based routing protocols extends Global Positioning System (GPS) services to route the packets in VANETs. 2.1. Topology-based routing Traditionally, topology-based routing protocols were initially proposed for MANETs, and were applied to VANETs because they have many common properties such as node mobility, distributed and self-organizing topology, non-existence of central control, etc. [25]. However, VANETs can be distinguished from MANETs because of their specific characteristics such as very high node mobility, limited degrees of freedom in mobility patterns which can be somewhat predictable, since vehicles move in rural or urban areas consisting of roads, highways, buildings, etc. In point of view routing, the MANET protocols are based on the IEEE 802.11 as a medium access control standard, and the transmission range is lower or equal to 250 m which is sufficient in such contexts where an important number of nodes tend to move with low speed. However, this transmission range is not enough for the

transmissions between vehicles because of their very high speed which made these transmissions instable. Consequently, topology-based routing protocols have been applied to VANETs but with the IEEE 802.11p standard which allows the transmission range to reach 300 m at least in order to make the network more stable [21]. Also, the routes used to disseminate data between vehicles have a short time of life compared with routes used by MANET nodes. This situation conducts to vehicular network partitioning [11]. All these differences have led to VANET researchers and designers adapting MANETs routing protocols to meet VANETs specificities [1,3]. Consequently, discovering routes in this case of topology-based routing, the setup of topological end-to-end paths between a source and a destination before sending the packets is the fundamental step. These topology-based routing protocols can be reactive or proactive. The most common MANET routing protocol that has been applied to VANET is the Ad hoc On-demand Distance Vector (AODV) [36] protocol. The route discovery method of AODV is based on routing tables which store the routes toward multiple destinations. Each destination is indicated using only the next hop node to reach this destination. The source disseminates a Route REQuest (RREQ) to its neighbors which in turn sends the same packet to their neighbors and so on, until the final destination is reached. Once the route request reaches the destination or an intermediate node which knows the path toward the destination, a Route REPlay (RREP) is sent back to the source node through the reverse route. AODV uses a sequence number to discover fresh paths and to prevent routing loops. Abedi et al. [1] extended AODV to apply it to VANET using directions and positions of the source node and the destination node obtained from GPS to find routes. Basically, source and destination directions are used for the next hop selection. To do this, an intermediate node can be selected as the next hop in the requested route if it is located and moves in same direction as the source and/or destination. This modified AODV routing protocol for VANET uses the mobility model of vehicles to support the various characteristics of VANETs. This reactive protocol establishes updated routes only when required. However, the intermediate nodes could indicate inconsistent routes if the sequence number is not updated and, the idea to choose the next hop in same direction of source and destination does not guarantee the optimality of the route found. In addition, the network can be flooded by multiple RREQ and RREP in addition to unnecessary bandwidth consumption due to periodic beaconing. Santa et al. [40] modified the well-known Optimized Link State Routing protocol (OLSR) proposed by Clausen and Jacquet [15] to apply it to VANET. Santa et al. proposed a unicast, link state, and proactive VANET routing protocol. With their routing protocol, each vehicle periodically sends HELLO messages in its transmission range to detect neighboring vehicles. Then, Topology Control messages (TC) are used to disseminate this information throughout the whole network. Moreover, TC messages are used to compute next hop destinations for all nodes in the network using the shortest hop forwarding paths. With the adapted OLSR algorithm, each vehicle selects MultiPoint Relays (MPRs) among all one-hop terminals, assuring that all neighboring vehicles which are two hops away can be reached through a minimum set of vehicles. Thus, only the MPRs vehicles forward messages leading to a decrease in VANET routing overheads when the network is highly dense. Since OLSR vehicles might have access to other networks such as the Internet or via an Ethernet link that are not running the OLSR protocol, a particular message called the Host and Network Association (HNA) message is periodically transmitted by this interconnected vehicle to inform other OLSR vehicles about this new interconnection and its parameters. In addition, this proactive protocol finds the different paths to all nodes, even if some paths are not requested, thereby reducing the route discovery delay. How-

Please cite this article in press as: S. Bitam et al., HyBR: A Hybrid Bio-inspired Bee swarm Routing protocol for safety applications in Vehicular Ad hoc NETworks (VANETs), J. Syst. Architect. (2013), http://dx.doi.org/10.1016/j.sysarc.2013.04.004

S. Bitam et al. / Journal of Systems Architecture xxx (2013) xxx–xxx

ever, the generated routing overhead is high because of the periodic messages exchanged in the network. Moreover, this protocol discovers many paths which are not needed. Namboodiri and Gao [31] proposed a Prediction Based Routing (PBR) protocol which is adequate for the mobile gateway scenario and predicts how long routes of vehicles on highways will last, and then it preemptively creates new routes to replace old ones before current ones fail. This protocol can reduce the routing overhead. As a reactive protocol, PBR uses the same basic operation (broadcasting route request and responding with route reply) for creating routes, and uses a mobility model of vehicles moving on a freeflowing highway. The authors studied the vehicle densities of Internet gateways to achieve sufficient connectivity to meet application requirements. The performance of PBR was demonstrated by simulation tests in which route failures were reduced after comparing with other protocols that do not use pre-emptive routing. Nevertheless, PBR is more constrained for applications which require minimal gateway switching. PBR also places additional load on gateways. Another topology-based routing protocol was proposed by Feng et al. [16]. This protocol consists of two algorithms: the VelocityAided Routing (VAR) algorithm was used as a packet forwarding process which is determined according to the relative velocity between the intended forwarding node and the destination node. The second algorithm called Predictive Mobility and Location-Aware Routing (PMLAR) algorithm was proposed to improve the routing performance by incorporating the predictive moving behaviors of mobile nodes in the protocol design. This protocol determines a region that a packet can be forwarded after predicting the future trajectory of the destination using the Gauss–Markov Mobility (GMM) model [27]. Also, PMLAR is based on proactive maintenance of routing paths instead of rediscovering them to enhance the routing performance. After simulations under different network topologies, the results obtained showed that the PMLAR protocol improves packet delivery ratio, end-to-end delay and control packet overhead. However, since PMLAR’s main idea is based on the GMM model, PMLAR can fail to predict the motion of each vehicle because GMM is a temporal dependency model which can be affected by the vehicle acceleration speed and direction changes in the geographical area. Nzouonta et al. [35] designed and implemented two protocols that belong to a class of routing protocols called Road-Based using Vehicular Traffic (RBVT) routing, named Reactive protocol (RBVTR) and Proactive protocol (RBVT-P). They aim to leverage real-time vehicular traffic information to set up road-based paths consisting of several road intersections that have, with high probability, network connectivity among them. RBVT-R discovers routes only when needed and, creates paths consisting of a set of connected road segments where each segment is formed with enough vehicular traffic. These paths are obtained using route discovery packets flooded from the source node to its neighbors and so on until they reach the destination. It is worth noting that RBVT-R uses geographical forwarding between intersections through the available vehicles in the path. RBVT-P periodically discovers all paths and save them on a graph which is used to transmit data after finding the shortest one. This process is performed using control packets named Connectivity Packets (CPs) which are sent in unicast mode in the network. CPs are periodically generated by a random number of nodes to discover the route and, traverse a road map path which is derived from a Depth-First Search (DFS) graph until the CPs reach the destination. Therefore, the source node computes the shortest path to the destination. RBVT-P switches to geographical routing in order to maintain a broken route. Their simulation results showed that the RBVT-R performed best in terms of data delivery compared with AODV and GSR [28]. In contrast, RBVT-P is better compared to other protocols in terms of the average

3

end-to-end delay. However, due to the periodic exchange of messages between vehicles, these protocols generate more control packets and flooded the network to obtain periodic up-to-date real time knowledge of the network connectivity [38]. In Bitam and Mellouk [8] a topology-based bee swarm protocol called Quality of Service Bee swarm routing protocol for VANET (QoSBeeVANET) was proposed. It is a reactive, multipath, QoS routing protocol designed for VANETs. Its main idea was inspired by the biological paradigm of communication between bees when searching for food. This protocol searches routes from the source to the destination using a request route packet called forward scout. This control packet is created and cloned by the source node to be launched to a limited number of nodes in the neighborhood (transmission range area) of the transmitter. In other words, only a percentage of source neighbors received the forward scout. The authors called this mechanism stochastic broadcasting. Each neighbor regenerates clones and sends the forward scout and so on until it reaches the destination or encounters an intermediate node that knows the path to the destination. Then, a backward scout is created and disseminated toward the source node through the reverse route. It is worth noting that this protocol guarantees QoS requirements such as end-to-end delay, throughput, and the packet delivery ratio. However, QoSBeeVANET can flood the network by control packets especially if any link fails. In Al-Rabayah and Malaney [2], the authors proposed a new routing protocol called Hybrid Location-based Ad-hoc Routing (HLAR) protocol that combines a modified AODV protocol with a greedy-forwarding geographic routing protocol. HLAR is a reactive protocol in which the source includes its location coordinates and those of the destination vehicle in a Route REQuest (RREQ) as in AODV. If the RREQ packet reaches the destination vehicle, the destination replies with a Route REPly (RREP). The intermediate vehicles are allowed to locally repair broken routes through a Route RePair (RRP) packet instead of just reporting a broken route to the source vehicle. RRP packet repairs link failures caused by vehicle mobility potentially leading to an increase in the routing overhead and degradation in network scalability. The simulation results obtained with HLAR demonstrated good scalability and optimal overhead even in the presence of high location errors. However, HLAR is very close to AODV and the geographic aspect can be better exploited to reduce the average end-to-end delay. Table 1 summarizes the properties of the topology-based routing protocols discussed above. 2.2. Geography-based routing Geography-based routing protocols have also been applied to VANET. They are also called position-based routing protocols in which the node positions are used to route data between vehicles. They perform a recovery strategy to overcome the void case when there is no routing progress based on nodes’ position data. A strong feature of these protocols is that the packets are routed to the destination without the knowledge of the network topology or a prior route discovery. In contrast, the source should determine its own position in addition to the position of the destination. An early work in this category is the Distance Routing Effect Algorithm for Mobility (DREAM) protocol which was proposed in [5]. It uses two basic ideas. The first one exploits the location information collected using Global Positioning System (GPS) technology and stored in routing table in order to send the packet to the destination or a neighbor node if available. Otherwise, a recovery procedure is executed. This transmission is performed before the time-out expiration. The second idea selects a route node by node until the destination is reached. Each node is chosen using a probabilistic method using locations and speeds of receivers. We note that the DREAM nodes periodically broadcast their physical loca-

Please cite this article in press as: S. Bitam et al., HyBR: A Hybrid Bio-inspired Bee swarm Routing protocol for safety applications in Vehicular Ad hoc NETworks (VANETs), J. Syst. Architect. (2013), http://dx.doi.org/10.1016/j.sysarc.2013.04.004

4

S. Bitam et al. / Journal of Systems Architecture xxx (2013) xxx–xxx

Table 1 Main benefits and drawbacks of topology-based routing protocols applied to VANETs (Notation: ++: more suitable, +: suitable, : less suitable, : unsuitable). Type

Topologybased

Routing protocol

(AODV) [36] (Enhanced AODV) [1] (OLSR) [15] (OLSR-VANET) [40] (PBR) [31] [16] [35] (QoSBeeVANET) [8] (HLAR) [2]

Mobility model VANET environment

Is the mobility model realistic?

Urban Urban Urban Urban/ highways Highway Urban Urban Urban Highway/grid

No No No No No No Yes Yes No

Route Stability

Route reliability

Performance characteristics

Longer Routes

Loops

Accuracy of destination found (Geographic position)

Delay

Packet Delivery Ratio

Routing Overhead

   

+ + + +

+ + + +

+ + + +

   

++ ++ + +

   

    

+ + + + +

+ + + + +

+ ++ + + +

    

+ + ++ ++ +

    

tion in their transmission range where nearby nodes are updated more frequently than the nodes far away. Simulation results obtained with DREAM have shown a significant reduction in endto-end delay after comparison with Dynamic Source Routing (DSR) [22]. Nevertheless, Dream is more sensitive to traffic load which makes it less suitable for VANETs where the traffic load varies significantly [4]. One of the most commonly used geographic-based protocols is the Greedy Perimeter Stateless Routing (GPSR) [23] proposed for wireless networks. It consists of two methods: the Greedy Forwarding method which is used wherever the forwarding of packets is possible, otherwise, the Perimeter Forwarding method is invoked. To achieve these goals, GPSR uses the positions of vehicles in its transmission range, and the destination to make its packet forwarding decision. In the case of greedy forwarding, the transmitter node chooses the optimal neighbor as the next hop which is the closest geographic node to the destination selected in a greedy manner. In other words, based on the neighbors’ positions, the transmitter selects the closest neighbor as its local optimal choice. It will be considered as the next hop to the packet’s destination. GPSR also uses a beaconing process to update its neighbors’ data (such as positions, etc.). If there is no intersection between the transmitter node and the destination node, the perimeter forwarding method is executed. It is based on the right hand rule in which, each node forwards packet through the perimeter to its first neighbor counterclockwise about itself. It is worth pointing out that under frequent topology changes resulting from the high mobility of vehicles, GPSR can use the local topology information to find the correct new routes quickly. This protocol was simulated over a full IEEE 802.11 and was compared with DSR in terms of routing overhead and the number of data packets delivered. The results showed GPSR’s scalability on densely deployed wireless networks. However, its greedy forwarding algorithm can fail if an interior node does not possess a neighbor in 2P/3 angular sector [23]. In addition, the perimeter forwarding algorithm finds a non-optimal route from the source to the destination which takes a longer time and is less efficient. In Lochert et al. [29] the Greedy Perimeter Coordinator Routing (GPCR) protocol has been proposed for VANETs routing. GPCR is based on the topology of real world streets and junctions to take advantage of the fact that streets and intersections are considered as a natural planar graph with the absence of any prior global information such as a static street map. It is composed of two procedures: a restricted greedy forwarding procedure and a repair strategy. In the first part, a special greedy forwarding is used to disseminate data to the destination. It routes the packets along streets

to cope with the area’s obstacles, and makes the routing decision at the junction nodes (called coordinators). The coordinator makes its decision about the street that the packet should follow in a greedy way where the neighbor with the largest distance to the forwarding node is selected. The second part called repair strategy, addresses the risk of a packet getting stuck in a local optimum and ensures the discovery of a better route than that found by the greedy forwarding, and which was considered as the best one. This repair strategy allows the junctions to take the appropriate routing decision and performs greedy routing between junctions. GPCR was simulated and obtained good results in terms of packet delivery ratio and the number of hops against GPSR. However, GPCR can fail to detect junction nodes which can be very distant and do not belong to the vehicle’s transmission range especially, in the rural area, along curve roads and long highways. In these areas, GPCR cannot make a routing decision because the GPCR decision policy is strictly dependent on junction nodes. Naumov and Gross [32] presented a geography-based routing scheme called Connectivity-Aware Routing (CAR) aimed at InterVehicle Communication in a city and highways. This protocol finds connected routes between the source and the destination after finding the destination’s position using nodes located in the Preferred Group Broadcasting (PGB) area. We note that the PGB area is a special region that belongs to the transmission range of a node which is limited by an inferior perimeter and a superior perimeter (calculated using the power of a signal that corresponds to the maximum transmission range and reduced by two configurable parameters to limit the transmission range between an inferior and superior adjusted values). Moreover, without a new discovery process, the obtained routes are auto-adjusted on the fly with the help of standing guards and traveling guards which are geographic temporary messages, and these guards use information necessary for routes maintenance. A standing guard will be activated to maintain the route if the source or the destination moves but, a traveling guard is activated if the source or the destination changes direction against the direction of the communication from the source to the destination. Then, the guards are buffered and passed from one node to another to propagate the new geographic information. CAR uses the beaconing procedure at the neighborhood level via Hello messages to update the geographic information in the transmission range of each node. CAR was simulated using ns-2 for city and highway scenarios, and was compared with GPSR. The results demonstrate that CAR delivers a clear improvement in the data delivery rate and the average data packet delay. Nevertheless, during the path discovery phase a significant routing

Please cite this article in press as: S. Bitam et al., HyBR: A Hybrid Bio-inspired Bee swarm Routing protocol for safety applications in Vehicular Ad hoc NETworks (VANETs), J. Syst. Architect. (2013), http://dx.doi.org/10.1016/j.sysarc.2013.04.004

S. Bitam et al. / Journal of Systems Architecture xxx (2013) xxx–xxx

overhead is incurred. Moreover, unnecessary nodes can be selected as part of the route which can affect the performance of the discovered route. In Zhao and Cao [45] several Vehicle-Assisted Data Delivery (VADD) protocols were proposed to forward the packet to the best road with the lowest data-delivery delay. Using predictable vehicle mobility, the authors adopted the idea of carry and forward, where a moving vehicle carries a packet until a new vehicle moves into its neighborhood and forwards the packet. By finding the next road to forward the packet using the existing traffic pattern, the delay is decreased. The experimental results showed that the proposed VADD protocols outperform DSR and GPSR in terms of packet delivery ratio, and routing overhead. It is worth noting that due to changes in traffic density and network topology, VADD can suffer from a significant end-to-end delay when sending packets. A Connectivity-aware Minimum-delay Geographic Routing (CMGR) protocol for VANETs was proposed in [42]. It considers two states of the network. When the network is sparse, it maximizes the chance of reception before packets expire, by taking the connectivity of streets into consideration. When the network is dense and consequently connected in most parts, it minimizes the delay by selecting non-congested routes that have a sufficient level of connectivity over time. It includes a target tracking mechanism to deal with the movement of target vehicles. CMGR uses a decision scheme for forwarding packets in sparse junctions which aims to minimize end-to-end delays by taking every promising forwarding opportunity into account. This work has conducted an experimental study starting with a mobility model of vehicles generated with microscopic street traffic simulation package (SUMO). The MAC/PHY protocol used is IEEE 802.11 that limits the transmission range to 250 m. This study has shown that CMGR presented significant improvements in terms of packet delivery ratio and dropped data packets. However, CMGR did not define a repair strategy to cope with significant packet losses which can be recorded. Moreover, its use is more adequate for urban scenarios and it is not very suitable for highway scenarios as in VANETs. To predict the future behavior of vehicles and to select a route with the longest lifetime to connect to the wired network, Benslimane et al. [6] introduced a routing protocol for VANETs. The proposed protocol aims at spreading the advertisement messages through multiple hops without flooding the network. To ensure connection continuity, this proposal assumes that the vehicles should seamlessly hand over connections to a new gateway before the current one terminate. This process is called seamless handovers. It can be considered as a new gateway discovery approach which creates a relatively robust network, and make the handovers seamless in order to reduce the routing overhead. This process restricts broadcasts to a predefined geographical zone, while allowing only some relays to rebroadcast the advertised messages. We note that stability metrics such as speed, direction, and location of vehicles are used to predict the future location of vehicles, and the period during which they stay in the transmission range of each other. The gateway discovery process uses a list of routes to different gateways that allow a vehicle to hand over the connection to the next available gateway before the current connection fails. If a vehicle does not receive advertisement messages, it should start sending out solicitation messages to find a new gateway. The performance of the protocol was compared with Greedy Perimeter Stateless Routing GPSR [12] and AODV+ [24] as an AODV version applied for wired networks. The simulation results show that the proposed scheme increases the packet delivery ratio and decreases the packet delay. However, this protocol is more appropriate for the highway scenarios compared to city-based scenarios which present the most important challenge with geographical routing for VANETs. In addition, the used relays could get overloaded with requests from other vehicles whenever the network is high dense.

5

Saleet et al. [39] proposed a routing protocol called the Intersection-based Geographical Routing Protocol (IGRP). It is mostly used for packet forwarding from vehicles to the Internet through a gateway by an effective selection of road intersections which a packet must traverse. IGRP tries to discover a path between the source and the destination which is consists of successive interconnected road intersections. These road intersections are selected in a way to maximize the probability of connectivity among the intersections while satisfying QoS requirements on tolerable delay, bandwidth, and error rate. In contrast, if the source node needs to exchange Internet data packets, it launches a route discovery to reach the Internet gateway which periodically updates and broadcasts the view of the local geographic network topology (all nodes geographic coordinates) to all nodes. Hence, each node can reach and transmit packets to any other node or can connect to the Internet through the interconnected road intersections formed on the basis of their geographic coordinates. This transmitting approach is called a geographic process. The comparative evaluation of IGRP with GPSR, GPCR and OLSR showed a significant VANET performance improvement in terms of transmission delay and bit error rate. We note that this protocol is not suitable for curve roads and long highways. In addition, IGRP assumes the use of the gateway to ensure the packet routing which is not always present especially, in the rural scenario. This protocol also incurs high routing overheads whenever the gateway updates the local network topology. In Bernsen and Manivannan [7] a position-based routing protocol for VANET called Reliable Inter-VEhiclar Routing RIVER was proposed, which actively monitors the traffic in real-time fashion. To do that, RIVER sends a probe message along streets without packet broadcasting. To transmit data packets, this protocol allows each node to identify the geographic coordinates of its neighbors using beacon messages. Furthermore, RIVER requires a basic knowledge of the physical location of streets ant their intersections to be used to forward data packets through an anchored routes. We note that the route consists of a set of anchor points in which two consecutive points are represented by a street edge marked by the geographic locations of its vertices. Then, the packet travels along the anchored streets which are more reliable but without using greedy forwarding from the source to the destination. RIVER was simulated using the IEEE 802.11 standard, and the Manhattan mobility model. The results obtained showed that RIVER yields a good throughput and improved reliable distribution of street graph information, especially for average to high-density traffic. However, RIVER can incur more delay to transmit data packet when the path found is not the shortest one. Moreover, this protocol requires streets and their intersections geographic coordinates to discover paths which is not suitable for non-city-based scenarios. In Wang and Lin [43], the authors proposed a Passive Clustering Aided Routing protocol, called Passive Clustering Aided Routing (PassCAR) protocol, to enhance the routing performance in oneway multi-lane highway scenario. It includes route discovery, route establishment, and data transmission phases. PassCAR selects suitable nodes to become cluster heads or gateways, which then forward route request packets during the route discovery phase. The suitability of nodes is ensured by a multi-metric election strategy that considers link reliability, link stability, and link sustainability as the main factors and quantifies them using the metrics such as node degree, expected transmission count, and link lifetime, respectively. Once the route is discovered, the destination node replies with a route reply packet to the source node, followed by data transmission through the routing path established. Performance evaluations have been performed using a VANET based on the IEEE 802.11 standard with a maximum transmission range equal to 250 m. Simulation results confirmed that PassCAR achieves a satisfactory path discovery ratio, network throughput,

Please cite this article in press as: S. Bitam et al., HyBR: A Hybrid Bio-inspired Bee swarm Routing protocol for safety applications in Vehicular Ad hoc NETworks (VANETs), J. Syst. Architect. (2013), http://dx.doi.org/10.1016/j.sysarc.2013.04.004

6

S. Bitam et al. / Journal of Systems Architecture xxx (2013) xxx–xxx

Table 2 Main benefits and drawbacks of geography-based routing protocols applied to VANETs (Notation. ++: more suitable, +: suitable, : less suitable, : unsuitable). Type

Geography-based

Routing Protocol

Mobility Model

(DREAM) [5] (GPSR) [23] (GPCR) [29] (CAR) [32] (VADD) [45] (CMGR) [42] [6] (IGRP) [39] (RIVER) [7] (PassCAR) [43]

VANET Environment

Is the mobility model realistic?

Grid Unobstructed plane Urban Urban/highway Urban Urban Highway Urban Urban Highway

No No No Yes Yes No No No Yes Yes

Route Stability

Route Reliability Longer Routes

Loops

Accuracy of destination found (Geographic position)

Delay

Packet Delivery Ratio

Routing Overhead

++ ++ ++ ++ ++ ++ ++ ++ ++ ++

         

         

         

         

+ + ++ ++ ++ ++ ++ + + +

+ ++ ++  + + + + + +

and path lifetime. Nevertheless, this protocol is less suitable for urban scenario and can lead to a considerable routing overhead due to its broadcasting policy to discover routes. Table 2 summarizes the main characteristics of recently proposed geography-based routing protocols for VANET. 2.3. Comparative study between topology-based and geography-based routing protocols In this section, the drawbacks of topology-based and geography-based routing protocols when applied to VANETs are summarized in Table 3. The main drawback of topology-based routing protocols is their route instability. This is because an established route consists of a set of nodes between the source and the destination that are affected by frequent broken routes in the presence of high vehicle mobility. The second drawback is the high routing overhead shared between nodes before transmitting data. It is due to the beaconing and Hello messages used to discover routes, to confirm this discovery and to maintain the paths found. Another limitation of topology-based category is the discovery of routes with a high latency and high transmission delay especially when the network is less dense. These delays are affected by route maintenance operations and repairs when the nodes are mobile. Consequently, an update of the discovered routes is required which causes an important delay. Topology-based routing protocols suffer from the problem of dropping packets also caused by the dynamic nature of the VANET environment. In this case, the old and unreliable routes are used to transmit data which will be dropped. On the other hand, despite the stability of the discovered routes, geography-based routing protocols suffer from several serious drawbacks. The first drawback is the difficulty in finding an optimal next hop node when searching the destination node, especially in the city-based scenario. In other words, the geography-based routing approaches can select the longest path through

Table 3 Summary of drawbacks of both Topology-based and Geography-based routing. Topology-based routing

Geography-based routing

1. 2. 3. 4. 5.

1. 2. 3. 4. 5.

Inadequate for rural scenarios Delayed transmissions Increased routing overhead Frequent broken routes More dropped packets

Inadequate for urban scenarios Using a longer path to transmit data Inaccurate GPS node coordinates Occurrence of inherent loops Frequent network partitioning

Performance Characteristics

the next hop in terms of geographic distance instead of the shortest one between the current node and the destination. This disadvantage is due to the lack of direct communication between two consecutive nodes due to obstacles caused by buildings, trees, etc. [28,26]. The second drawback results when the recovery strategies of the geography-based routing are applied. As we mentioned previously the recovery strategies are based on planar graph traversals which often require no cross-links in order to recover routes. This requirement is hard to achieve in practice for VANETs because of the presence of radio obstacles and the high mobility of vehicles to ensure there are no intersections between links. This can be easier in an environment characterized by a uniformly distributed instead of the planar graphs [29]. The third disadvantage of geography-based routing protocols is the inherent loops caused by the vehicle’s mobility and its strict positions when discovering or maintaining routes. These loops can lead to loss of the ability to memorize past traffic history which can help to prevent launching a new routes discovery [17]. Another disadvantage of geography-based routing algorithms is the use of GPS device which can fail because of various reasons such as the presence of obstacles or the atmospheric conditions which could block the GPS signal. Geographic measurements with commercial GPS receivers showed errors in the reporting of GPS positions [41], and sometimes packets may get forwarded to the wrong direction causing higher delays or even network partitions. Furthermore, these geography-based routing protocols do not consider real-time positions of vehicles in their decision-making procedure and consider only static roadmap data [20]. Depending on the brand of GPS, updating these static roadmaps data is expensive. This situation often leads to a wrong decision. To address the drawbacks of topology-based and geographybased routing approaches, we propose the design of a hybrid routing protocol called Hybrid Bee swarm Routing (HyBR) protocol for VANET. Our proposed protocol is a unicast and a multipath routing protocol which guarantees requirements of VANET safety applications, placed as stringent requirements such as end-to-end delay, packet delivery ratio and normalized overhead load. As we demonstrate below HyBR can provide stable and reliable routes between the source node and the destination node with an optimal distance, reduced delay, increased packet delivery ratio and low routing overheads. HyBR combines two fundamental routing methods namely, topology-based routing and the geography-based routing in order to reap their benefits on one hand and avoid their drawbacks at the same time.

Please cite this article in press as: S. Bitam et al., HyBR: A Hybrid Bio-inspired Bee swarm Routing protocol for safety applications in Vehicular Ad hoc NETworks (VANETs), J. Syst. Architect. (2013), http://dx.doi.org/10.1016/j.sysarc.2013.04.004

7

S. Bitam et al. / Journal of Systems Architecture xxx (2013) xxx–xxx

3. Hybrid Bee swarm Routing (HyBR) protocol

HyBR protocol

In this section, we describe our proposed bio-inspired Hybrid Bee swarm Routing (HyBR) protocol (a unicast routing protocol) for VANETs.

Topology-based routing (VANET high density)

Geography-based routing (VANET low density)

Fig. 1. HyBR routing protocol procedures based on VANET density.

3.1. Bee in nature 3.1.1. Bee components and communication The bee (Apis mellifera) is a social and domestic insect native to Europe and Africa. There are between 60,000 and 80,000 living elements in the hive. The bees feed on nectar as a source of energy in their lives and use pollen as a source of protein in the rearing larvae. Generally, the bee colony contains a single breeding female known as the Queen, a few thousands of males called Drones, several thousands of sterile females called Workers, and many young bee larvae called Broods. The bees share a communication language of extreme precision, based on the dances which are performed by the workers (called ‘‘scouts’’) in this exploration phase. After finding food and returning to the hive, the scout informs others about the distance, direction, quantity and quality of food found. With their visual, tactile and olfactory perception, the other bees perceive the transmitted information. There are two types of dances: the round dance when food is very close. This dance indicates only the direction; the second type of dance is the waggle dance in which the bee effects a repeated movement that forms a drawing like the number eight. This scheme indicates the distance and the direction of the food source. The distance between the food source and the hive is transmitted depending on the speed of the dance. If the dance is faster, then the food distance is smaller. The direction (the angle between the food source and the sun relative to the hive) is shown by the inclination of the dance vertically with an accuracy of ±3°. The food nature is indicated by the odor of the bee when it is rubbed. The food amount depends on the wriggling of the bee. The more the wriggling, the higher is the quantity. Often, the bees’ communication is shared when they search food sources, starting with some bees called ‘‘scouts’’ which navigate and explore the region aiming to find a food source. In the positive case, they return to the beehive at a level named ‘‘dance floor’’ to transmit and share this discovery with the others through round or waggle dances according to the discovery distance, quality, etc. Thus, some bees are recruited and become foragers which increase with the proportion of food found. This step is known as the exploration phase which is followed by the exploitation, in which the forager collects food and calculates the quantity to make a new decision. The forager can continue to collect food by remembering the best location, or it leaves the source and returns to the beehive as a simple bee [9].

High density sub-network (α ≥ TR/β)

3.1.2. Applying bees’ communication principle to VANET Inspired by communication among bees when they search their food source, the VANET environment can be viewed as a bees’ environment. The end-point sender which could be either a vehicle or a roadside base station corresponds to the beehive, and the destination which may be a vehicle or a roadside base station called hereafter a node, corresponds to the bees’ food source. Intermediate vehicles or roadside base stations are represented by workers. 3.2. Basic principles of HyBR As mentioned above, HyBR is a hybrid protocol which applies a topology-based routing approach when the network density is high (e.g., city-based VANET) and applies a geography-based routing approach when the network is not dense (e.g., highways) as shown in Fig. 1. Using GPS devices, outdoors or through other means, each node saves the position information of all VANET nodes in a table called a positions table which is updated whenever the network topology changes. Moreover, each node possesses its own routing table which contains the various routes toward the desired destination. Only the next hop toward the destination is indicated.  Partitioning of the network into sub-networks to perform the appropriate routing: Network density is used to determine the type of routing method to use in the VANET environment. This decision is made by the source node after estimating the VANET density based on the number of all nodes extracted from the positions table. In other words, using its positions table, the source node checks the network between the source and the destination after dividing it on a set of sub-networks where each one has a perimeter equal to the transmission range (Fig. 2). If the number of nodes in a sub-network is superior to a threshold ‘a’ called density coefficient, calculated using (1), the topologybased routing is applied. Otherwise, the geography-based routing is invoked. a represents the number of vehicles in the checked sub-area.

a¼

TR b

ð1Þ

Low density sub-network (α < TR/β)

: Source node : Intermediate node : Border node : Destination node Fig. 2. HyBR transmission based on the network density.

Please cite this article in press as: S. Bitam et al., HyBR: A Hybrid Bio-inspired Bee swarm Routing protocol for safety applications in Vehicular Ad hoc NETworks (VANETs), J. Syst. Architect. (2013), http://dx.doi.org/10.1016/j.sysarc.2013.04.004

8

S. Bitam et al. / Journal of Systems Architecture xxx (2013) xxx–xxx

where, TR is the transmission range perimeter, and b is an observer parameter that specifies the density of the transmission range. In this work, we have chosen b = 6 as an empirical value which leads to the best results reached after a set of tests. Therefore, if TR = 300 m, a = 300/6 = 50 vehicles. In this case, we consider that this sub-area is dense. After, the source node calculates the sub-routes using topological routing of its routing table, and also, finds the sub-routes used to disseminate packets using geography-based routing using its positions table. All these sub-routes are put in the packet header in sorted manner (i.e., the sub-routes are put to indicate the whole route from the source to the destination) in order to send all data packets. Fig. 3 illustrates two consecutive sub-paths that constitute the entire path between the source and the destination. The first sub-path is reached using the topology-based procedure, and the second one is obtained by the geography-based procedure. These two sub-paths are placed by the source node at the header of each data packet which is sent out to the destination. It is worth noting that if the network is divided in sub-networks where each one uses the appropriate routing based on the sub-network density, a border node is defined by the source node between each two consecutive sub-networks. This border node plays the role of an intermediate destination of the first sub-network, and as an intermediate source for the second one. 3.2.1. Topology-based routing procedure of HyBR Topology-based routing is a reactive routing method that is applied if the network is highly dense. This procedure consists of three phases: (a) the beaconing phase which ensures the local updating of the node links in its transmission range of the current node, (b) the route discovery phase used to find different routes from the source to the destination, and (c) the route repairing phase which is performed when the links are broken. 3.2.1.1. Packet types used in the topology-based procedure. In this procedure, HyBR uses two types of packets: Scout and Forager. In order to update the immediate neighbor list, it frequently employs a beacon packet. It is worth noting that a node is considered to be an immediate neighbor of the current node if it is within its transmission range. For the route repairing process, an error scout packet is used to warn the nodes involved and it is launched when a link is broken. Scout: It is a control packet used to discover the route. If the destination route is unknown, the scout is called a ‘Forward scout’ and, is sent from the beehive (source node) as a route request to find route toward the food (destination node). When the food is found, it returns to the beehive as a backward scout to transmit this discovery information to its nest mates (foragers or data packets).

High density sub-network

(i) Forward scout: The forward scout packet is created when the source node needs to transmit data to an unknown destination after consulting the routing table. It is cloned and launched to nodes in the neighborhood using a stochastic broadcasting until it reaches the destination. It remembers its route to the destination using the routing tables of the visited nodes to enable a return path to the source. The forward scout (shown in Fig. 4) consists of a scout identifier which is unique and is identified by an integer number. After the creation of each forward scout, its scout identifier is incremented. Also, the forward scout consists of a beehive identifier which is the source node address. The combination of scout and beehive identifiers is used to identify the route request and it helps to prevent duplicate route requests. This control packet includes the food identifier (the destination node), the minimum bandwidth requested and the maximum delay allowed by the source node, and a lifespan field used to limit the number of hops by the forward scout and guarantee loopfree route requests. If the lifespan is reached, the packet will be dropped. If beehive does not receive any response after the waiting time expiration, the lifespan field will be increased. The forward scout also contains a hop count field which is the number of hops from the source to the node handling this scout. Moreover, the scout packet includes a timestamp field which records the transmission time used to evaluate the route to the destination found. (ii) Backward scout: A backward scout is generated by the destination or an intermediate node that knows the destination once the route to the destination is found. The backward scout is propagated to the source node along the reverse path. It consists of a scout identifier, the beehive identifier, the food identifier, as well as a hop count field to indicate the number of hops from the source node to the destination. The hop count field is filled with the hop count of the forward scout when finding the requested route. Moreover, the backward scout includes the minimum bandwidth requested and the maximum delay allowed by the source node, in addition to a timestamp field used to transfer the delay between the source and the destination. The backward scout uses the lifespan field value as a time to live value. Forager: It is a data packet which is used to transmit the data between nodes. If the route is unknown, the data packets are queued until the desired route is discovered, and then they will be launched to the destination with the data that needs to be transmitted. 3.2.1.2. Routing table used in the topology-based procedure. The routing table has a three-dimensional structure and is situated at the level of each node. It consists of a set of vectors. Each vector corresponds to one destination in the network that consists of a set of entries. Each entry represents a route to the destination. The avail-

Low density sub-network

: Source node : Node in the path : Border node : Destination node : sub-path discovered by topology-based procedure : sub-path discovered by geography-based procedure Fig. 3. HyBR route based on two consecutive sub-paths.

Please cite this article in press as: S. Bitam et al., HyBR: A Hybrid Bio-inspired Bee swarm Routing protocol for safety applications in Vehicular Ad hoc NETworks (VANETs), J. Syst. Architect. (2013), http://dx.doi.org/10.1016/j.sysarc.2013.04.004

S. Bitam et al. / Journal of Systems Architecture xxx (2013) xxx–xxx

9

Scout Id Beehive Id Food Id Bandwidth Delay Lifespan Hop count Timestamp

(a): Forward scout packet Field

Description

Scout Id

To identify the route request (forward scout)

Beehive Id

To identify the source node

Food Id

To identify the destination node

Bandwidth

The minimum bandwidth requested by the source node

Delay

The maximum delay allowed by the source node

Lifespan

To limit the number of hops traveled by the forward scout (it helps to guarantee loop-free)

Hop count

Number of hops traveled by the forward scout

Timestamp

To record the time traveled by the forward scout

(b): Table of forward scout fields Fig. 4. Structure of forward scout.

ability of more than one path for one destination can be used to transmit more data concurrently. Each entry (Fig. 5) includes the food identifier, the node identifier of the next hop node toward the destination and the identifier node of the prior hop towards the source node. It is worth noting that the next hop field is used in the transmission of packets whereas the prior hop field warns of a possible link update in the transmission range of the node. This prior hop helps a packet to reach the source node through the reverse route. Link updates can occur when either there is a connection link failure or there is a change in parameters (such as bandwidth and delay over the link) required by the road safety service. To detect the link updates, beacon packets are periodically sent in the transmission range of each node. Other fields such as the hop count field and the number of links separating the node from the destination are also stored in the routing table entry. The weight factor related to the route quality is also one of the fields of the entry indicating a route. It is used to express the delay incurred to reach

Node i

IP Destination 1 Route 1

Route k

IP Next hop

IP Next hop

IP Prior hop

IP Prior hop

Hop count

Hop count

Link weight

Link weight

ID Scout

ID Scout

Timeout

Timeout

Route delay

Route delay

Bandwidth

Bandwidth

IP Destination j

Fig. 5. Routing table of node i.

the destination. Consequently, the higher the weight associated with a route, the lower end-to-end delay between the source and the destination. Routing data carried with the forward scout are temporarily saved in the routing table entry when the forward scout executes the route request. Afterwards, it will be permanently recorded by the backward scout. Routing data help to determine the freshness of the route. Furthermore, a timeout field is used to retransmit a new forward scout if the last forward scout does not return after the waiting time expires. In addition, results of end-to-end delay and the estimated bandwidth between this node and the destination are also recorded. These QoS values are calculated from the value stored in the timestamp field of the forward scout packet. It is worth noting that the route is considered as valid only when the road safety guarantees are satisfied and there is no link failure over the route. 3.2.1.3. Phases of the topology-based procedure. Beaconing phase: This phase is carried out by each node in the VANET environment and aims to inform the node of its neighbors and update its active links. This phase also helps to estimate the road safety requirements such as the available bandwidth and the measured endto-end delay of each link. All nodes in the VANET periodically broadcast a beacon packet to the nodes located in their transmission range. When a neighbor node receives the beacon packet, all entries in its routing table about the sender will be considered as valid. The routing information in the routing table is marked as invalid if a node does not get information from the node’s neighbor for a specified amount of time. Consequently, the other nodes are informed of this unavailable link using an error packet. Route discovery phase: When one node is responsible to transmit data, it first checks whether the route is present in its routing table and the road safety requirements are satisfied. In other words, the required bandwidth and delay in the proposed route should be less than the available bandwidth and the actual delay respectively. If there are sufficient forgers, the source node transmits the data, otherwise, data is buffered until new foragers are recruited. To dis-

Please cite this article in press as: S. Bitam et al., HyBR: A Hybrid Bio-inspired Bee swarm Routing protocol for safety applications in Vehicular Ad hoc NETworks (VANETs), J. Syst. Architect. (2013), http://dx.doi.org/10.1016/j.sysarc.2013.04.004

10

S. Bitam et al. / Journal of Systems Architecture xxx (2013) xxx–xxx

cover a route, the source node generates and clones several forward scouts to be launched and broadcasted stochastically to their immediate neighbors. As an experimental value, we have fixed the threshold for the number of neighbors to 80% because it leads to optimal results [8]. All cloned forward scouts share the same scout identifier, the beehive identifier, the food identifier and the maximum lifespan of the route request. The hop count of a forward scout is initialized to zero and is increased every time it encounters a node until it reaches the destination. Whenever the forward scout encounters a node, the cumulative travel time is saved in the timestamp field. The visited node checks the road safety requirements followed by the route discovery process. Otherwise the forward scout will be dropped and the route discovery process is not executed. At the same time, each intermediate node checks if it has already received a forward scout with the same scout identifier and beehive identifier. If so, the forward scout will be dropped, else the node saves the forward scout identifier and the source node identifier at the prior hop field in the routing table entry. This saved data keep track of the visited node about the identity of the last (newer) forward scout. This step prevents further consideration of the same scout or of an old one. Afterwards, the route existence is checked with the intermediate node’s routing table. If the intermediate node does not know the route to the destination; it clones and rebroadcasts stochastically the forward scout like the source node. However, if the route is discovered in the routing table of this intermediate node, it generates and launches a backward scout to the source node along the reverse path using the prior hop field located in the routing table. Route repairing phase: When links are broken in VANET, HyBR performs the repair as follows. The periodic sending of beacon packets in the transmission range between the node and its neighbors helps to ensure the stability of the connections. If there is connectivity problem, the node detects the broken link and any route passing through this broken link is considered as disconnected. In addition, a road safety requirement violation can also be detected by the node if there is insufficient bandwidth or a large delay over the link. In these cases, an error scout is generated by the current node and is sent to the source node via the previous node. This step aims to inform any node in the reverse route about the broken link to remove it from its routing table and to restart a new route discovery by the source node if necessary.

destination’s position is out of the sender’s transmission range and causes routing to take another path through a neighbor node leading to a longer geometrical distance (as illustrated in Fig. 6). As result, the metaheuristic method provides the optimal route to use to transmit packets in with the least geometric distance from the source to the destination. In this work, we choose to use the Genetic Algorithm (GA) proposed by Holland [19] as the metaheuristic method which generated the optimal route in a decreased processing time. We note that each node in the VANET periodically updates the geographic positions of the network nodes using a GPS service.

3.2.2. Geography-based procedure of HyBR If the network density is low (i.e., the number of nodes in a subnetwork is lower than the density coefficient), HyBR uses a metaheuristic method to find the shortest paths between the source and the destination from the basic entries of the positions table. The routes found should guarantee the road safety requirements such as the acceptable end-to-end delay and the required bandwidth. This process is applied to all VANET topologies even in the case where the only route to the destination requires transmitting the packet temporarily along a longer geometrical distance from the destination [23]. This phenomenon occurs when the

popk ¼

3.2.2.1. Genetic algorithm for geographical routing in HyBR. In this section, the application of GA to discover the optimal route between the sender and the receiver in VANET is presented. First, the individual and population representations are described. We then present the GA initialization and the fitness function. Next, the selection of the parents, crossover and mutation operators are explained followed by the stopping criterion used for the iterations. Fig. 7 shows the pseudo-code of the GA. Individual and population representation: The individual of GA is represented by a simple route from the source to the destination through a set of intermediate nodes. The route is encoded using a string expressed by order on node route numbers [30]. Consequently, the population is represented by a vector of N individuals. Initialization and fitness function: As mentioned above, each node maintains a positions table that consists of a set of entries. Each entry determines the position of each other node in the network using a GPS. Therefore, the source could establish an initial route to the destination at random as an initial individual. By similarity, a set of individuals can be generated as an initial population. To evaluate an individual, an objective function (i.e., fitness function) is applied that considers the shortest path from the source to the destination as the best one. In other words, the fitness fi of the individual ‘i’ is the sum of the distance (dis) between each two adjacent nodes ‘nj’ and ‘nj+1’ in the path from the source node ‘s’ to the destination node ‘d’, calculated by the following formula:

fi ¼

d X disðnj þ njþ1 Þ j¼s

Thus, the population fitness ‘popk’ is the sum of the individuals’ fitness ‘fi’, calculated by the formula: N X fi i¼0

It is worth noting that the road safety requirements (the allowed end to end delay and the required bandwidth) are considered as two strict constraints to be checked by the fitness function. Selection, crossover and mutation: To determine two parents N/2 times, a selection strategy is chosen where N is the number of indi-

1. Initialize population at random 2. Evaluate fitness of population 3. While stopping criteria are not satisfied 4. Selection 5. Generate offsprings

: Source node : Intermediate node : Destination node : Closer path (but inutile) : Distant path (used for routing)

Fig. 6. Routing along distant path instead of a closer path (no recovery between source and destination).

6. Crossover 7. Mutation 8. Replacement 9. End while Fig. 7. HyBR genetic algorithm pseudo-code.

Please cite this article in press as: S. Bitam et al., HyBR: A Hybrid Bio-inspired Bee swarm Routing protocol for safety applications in Vehicular Ad hoc NETworks (VANETs), J. Syst. Architect. (2013), http://dx.doi.org/10.1016/j.sysarc.2013.04.004

S. Bitam et al. / Journal of Systems Architecture xxx (2013) xxx–xxx

viduals in the population. To do this, we use the Roulette Wheel Selection (RWS) presented in Goldberg and Deb [18] in which each parent is selected following a probability pi proportional to its fitness fi so this probability is calculated as follows:

fi pi ¼ PN

j¼1 fi

After, a binary operator called crossover is executed with a probability of pc to generate two offsprings. In this work, we use two-point crossover. Two-point crossover chooses for both selected parents, two intermediate nodes at random. It then combines the middle part of the first parent with the two extreme parts of the second one and vice versa. With probability of pm a mutation operator is performed on each of the new offsprings. It is a unary operator in which the offspring has the chance to be changed according to the pm probability. To mutate an offspring in this work, we choose from the selected offspring which is a path from the source to the destination, two intermediate nodes at random, to be permutated. This mutation operation is called a two-point mutation. Stopping criterion: The stopping criterion chosen is dynamic. This means that the number of iterations is constrained by a maximum threshold MaxIT of iterations only if a stagnation state is not reached. The stagnation state is a minimum threshold of MinIT or lower. MaxIT and MinIT are considered as user parameters, which are fixed by the user as empirical values defined following the experimental observations. 4. Experimental evaluation of proposed HyBR approach We evaluated our proposed hybrid routing protocol (HyBR) as for VANET using the network simulator ns-2 [33]. HyBR was evaluated using the following road safety metrics namely average endto-end delay, average bandwidth, Packet Delivery Ratio (PDR) and Normalized Overhead Load (NOL). The average end-to end delay is the average time taken between the generation of a packet by the source node and the time when this packet is received at the destination which includes all possible delays caused by buffering

11

during route discovery latency, queuing at the interface queue, retransmission delays at the MAC, and propagation. The Packet Delivery Ratio (PDR) metric is the number of packets successfully received by the destination node to the number of packets that was transmitted by the source nodes. The Normalized Overhead Load (NOL) represents the total number of routing packets divided by total number of data packets delivered. It is used to indicate the extra bandwidth consumed by the overhead to deliver data traffic. Moreover, we also compared the performance of our proposed HyBR routing protocol with AODV [36] and GPRS [23] corresponding to topology-based and geography-based protocols respectively.

4.1. A realistic mobility model used in our simulation tests During our review of past related works earlier, we noted that most experiments on VANET simulations have been based on the native form of the ns-2 mobility models or on theoretical mobility models which affected the results due to their unrealistic mobility of nodes. These theoretical mobility models do not capture real VANET parameters such as the city traffic density, the environment geography (highways, roads, obstacles, etc.), the vehicles’ velocity, positions, density and movements Camp et al. [13] and Aravind and Tahir [3]. In contrast, to evaluate our proposed routing protocol we used a realistic mobility model of VANETs which is based on urban traffic features such as the limited capacity of the roads and the existence of intersections, obstacles, buildings, vehicles density, movement, etc. The experimental area chosen is the Biskra city downtown located in the southeast of Algeria as shown in Fig. 8. This area consists of a set of boulevards, roads, streets and a set of junctions and intersections which are frequented by a large number of vehicles every day due to their presence in a strategic place especially during working hours. However, it is considered as low density area during other hours on weekdays or on weekends. These two kinds of density have been taken into account in our mobility model. In addition, the various mobility patterns of vehicles are also taken into consideration as well as the existence of traffic lights. We

Fig. 8. Area of experimentation: downtown of Biskra city (Algeria).

Please cite this article in press as: S. Bitam et al., HyBR: A Hybrid Bio-inspired Bee swarm Routing protocol for safety applications in Vehicular Ad hoc NETworks (VANETs), J. Syst. Architect. (2013), http://dx.doi.org/10.1016/j.sysarc.2013.04.004

12

S. Bitam et al. / Journal of Systems Architecture xxx (2013) xxx–xxx

RS V

: Roadside station : Vehicle

V V

V

RS

RS V RS

V

Fig. 9. Abstraction of the experimental area.

Fig. 10. Average end-to-end delay versus network density and improvement of HyBR over other protocols.

Fig. 11. Packet Delivery Ratio versus network density and improvement of HyBR over other protocols.

Please cite this article in press as: S. Bitam et al., HyBR: A Hybrid Bio-inspired Bee swarm Routing protocol for safety applications in Vehicular Ad hoc NETworks (VANETs), J. Syst. Architect. (2013), http://dx.doi.org/10.1016/j.sysarc.2013.04.004

13

S. Bitam et al. / Journal of Systems Architecture xxx (2013) xxx–xxx

Fig. 12. Normalized overhead load versus network density and improvement over other protocols.

Table 4 Main benefits of HyBR (Notation. ++: more suitable, +: suitable). Type

Hybrid protocol

Routing Protocol

Proposed Routing Protocol (HyBR)

Mobility Model VANET Environment

Is it realistic? (mobility model)

urban/ highway

Yes

Route Stability

Route Reliability Longer Routes

Loops

Accuracy of destination found (Geographic position)

Delay

Performance Characteristics Packet Delivery Ratio

Routing Overhead

++

+

+

+

++

++

+

varied the number of nodes (i.e., vehicles) from 20 to 50 nodes from a low density to a high density network, with a uniform distribution of vehicles. In this geographic area, ten (10) roadside base stations are made as RVC devices, and the abstraction of the experimental area is presented in Fig. 9. We note that in this work, three sets of experiments have been conducted. In the first one, 20 nodes have been used in the network (corresponding to low density). In the second set of experiments, the network consists of 35 nodes (corresponding to medium density). In the third set of experiments, we considered a network with 50 nodes (corresponding to high density). These 50 nodes are located when the simulation is launched over an area of 500  500 m2 from the whole area (the left part of the area) and then, they occupy the whole area. In the beginning of this third simulation, the topology-based procedure is applied, however, when nodes start moving, the geography-based procedure can be applied since the number of nodes is less than 50 in the left part area (500  500 m2). As cited above, we remind, that the threshold (a = 50) allows to switch from topology-based to geography-based procedure. In contrast, the two first experiments call only the geography-based procedure to send packets.

which they exchange packets using the IEEE 802.11p protocol. The frequency band used is fixed at 5.9 MHz with a transmission range of 300 m. The nodes exchange data packets using the Transport Control Protocol (TCP) where a packet has a size of 1000 bytes. The number of bytes transmitted per second is equal to 0.01 Mbytes. For each simulation run, the number of transmitter nodes is the same as the number of receiver nodes. Each node is characterized by an initial position and movement following the mobility model with a speed that varies from 0 to 20 m/s. When the metaheuristic method is called and if the network density is low, we choose the population size N equal to 10 individuals, the crossover probability pc is fixed at 95%, the mutation size pm is set to 10%, the stagnation state MinIT is set to 30 iterations and the maximum number of iterations MaxIt is set to 200. We have set these user parameters to these specific values after several experimentations which have led to the best results.

4.2. Simulation environment and parameter settings

As we mentioned previously, the performance of HyBR protocol has been analyzed using three road safety metrics namely: average end-to-end delay, packet delivery ratio, and normalized overload load. After running the simulation tests, we obtained results shown in Figs. 10–12.

As we mentioned earlier, HyBR is simulated using ns-2 version 2.35 set up on the Ubuntu 12.04 Linux operating system. All vehicles move in an area of 1000  1000 m2 during 500 s in

5. Performance results and discussion 5.1. Performance results

Please cite this article in press as: S. Bitam et al., HyBR: A Hybrid Bio-inspired Bee swarm Routing protocol for safety applications in Vehicular Ad hoc NETworks (VANETs), J. Syst. Architect. (2013), http://dx.doi.org/10.1016/j.sysarc.2013.04.004

14

S. Bitam et al. / Journal of Systems Architecture xxx (2013) xxx–xxx

5.2. Discussion Our simulation results show that HyBR outperforms AODV and GPSR in terms of average end-to-end delay (as shown in Fig. 10). It takes 0.08s in the low density case, 0.17s in the medium density case, and 0.29 when the network is highly dense, compared to AODV and GPSR which resulted in 0.22s and 0.90s in their best cases, respectively. HyBR improves the end-to-end delay from 5% to 61% against AODV, and from 69% to 112% compared to GPSR. A detailed analysis of this improvement reveals that our proposed HYBR protocol ensures packet exchanges with minimum delay due to its multipath connectivity. Moreover, when the network density is low, geographical routing is invoked by HyBR which helps to give the shortest route quickly without spending time in the route discovery process. When the network density is medium, both geographical and topological are applied after dividing the network into low-density or high-density sub-networks to perform the appropriate routing. Topological routing is used if there are many nodes in areas such as cities. Consequently, the main drawback of the geography-based routing which is the calculation of the next hop, is avoided. These performance results obtained for HyBR (0.08, 0.17 and 0.29 s), show its suitability to satisfy the road safety requirements. Fig. 11 depicts the packet delivery ratio values of the HyBR, AODV and GPRS protocols for various network diversities. The results show a high delivery of packets with HyBR: 99.49% if there are 20 nodes, 97.69% with 35 nodes and 98.79% when the network is dense which are slightly higher compared to AODV. AODV and HyBR can find the destination due to their discovery methods which are based on the RREQ between nodes in local area and, are not based only on the geographic information of nodes used by GPSR that can lead to inaccurate position of the destination. Therefore, the PDR is higher for AODV and HyBR. We note that compared to AODV, HyBR uses a local stochastic broadcasting to find routes which reduces the network congestion thereby improving the packet delivery ratio. However, GPRS achieved the worst results because of a high rate of dropped packets due to the inaccuracy in the destination found. This factor is considered to be one of the major drawbacks of the geographic routing protocols proposed for VANETs. Our proposal improves the PDR against GPSR from 57.79% to 92.29%. As for the normalized overhead load results obtained, we note that HyBR offers acceptable values 40.02%, 41.50% and 62.87% for different network densities. It is worth mentioning that GPSR gives the best results (38.39% to 47.86%) because it does not generate a route request packets and the maintenance mechanism is not used. These procedures require sharing more control packets. In contrast, AODV performs the worst especially if the network density is low or medium. In these two cases, HyBR performs better than AODV. However, AODV is better compared with HyBR if the network density is high because HyBR uses in this case both topology and geographical routing procedures which increase the number of control packets. We note that our proposed HyBR approach yields stable results close to GPSR and does not exceed 15.01% in the worst case. Based on the performance results presented above, we conclude that HyBR offers an important improvement in the average endto-end delay and packet delivery ratio with an acceptable routing overhead compared to AODV and GPSR. Finally, we summarize in Table 4 the main benefits of our proposed hybrid protocol (HyBR) in terms of the most important routing metrics for VANETs.

6. Conclusion In this paper, we have proposed a new hybrid routing protocol called HyBR for VANETs that is more suitable in the urban scenario

as well as for rural context. HyBR guarantees road safety service quality which is the most objective of transportation systems. HyBR uses two main procedures to deal with the varying density of VANET. When the network density is low a geography-based routing approach is performed otherwise, a topology-based routing protocol is executed. It is worth noting that this protocol can apply the two procedures at once based on the network density. In other words, the area is subdivided in sub-areas according to the network density, and then the appropriate routing procedure is executed. As result, the high quality information needed for VANET safety is correctly shared between network nodes. We have demonstrated, by a set of simulation tests and through a realistic propagation model that HyBR outperforms the standard topology-based routing protocol (AODV) and the conventional geography-based routing protocol (GPSR), in terms of average end-toend delay, and packet delivery ratio. Furthermore, it provides an acceptable normalized overhead load measure. As part of our future work, we plan to use HyBR protocol across hybrid networks to provide Internet access as well as cloud computing connections for ITS applications. Acknowledgements The authors would like to thank the anonymous reviewers for their valuable comments which help us to improve the content and presentation of this paper. References [1] O. Abedi, M. Fathy, J. Taghiloo, Enhancing AODV routing protocol using mobility parameters in VANET, in: IEEE/ACS International Conference on Computer Systems and Applications (AICCSA 2008), Doha, Qatar, 2008, pp. 229–235. [2] M. Al-Rabayah, R. Malaney, A new scalable hybrid routing protocol for VANETs, IEEE Transactions on Vehicular Technology 61 (6) (2012) 2625–2635. [3] A. Aravind, H. Tahir, Towards Modeling Realistic Mobility for Performance Evaluations in MANET, in: 9th International Conference on Ad Hoc Networks and Wireless (ADHOC-NOW’10), Edmonton, Alberta, Canada, 2010, pp. 109– 122. [4] M. Bakhouya, J. Gaber, M. Wack, Performance Evaluation of DREAM Protocol for Inter-vehicle Communication, in: 1st International Conference on Wireless Communication, Vehicular Technology, Information Theory and Aerospace & Electronic Systems Technology, (Wireless VITAE), Aalborg, Denmark, 2009, pp. 289–293. [5] S. Basagni, I. Chlamtac, V.R. Syrotiuk, B.A. Woodward, A distance routing effect algorithm for mobility (DREAM), in: 4th IEEE/ACM annual International Conference on Mobile Computing and Networking (MobiCom’98), Dallas, TX, 1998, pp. 76–84. [6] A. Benslimane, S. Barghi, C. Assi, An efficient routing protocol for connecting vehicular networks to the Internet, Pervasive and Mobile Computing, Elsevier 7 (1) (2011) 98–113. [7] J. Bernsen, D. Manivannan, RIVER: a reliable inter-vehicular routing protocol for Vehicular Ad hoc Networks, Computer Networks, Elsevier, in press. [8] S. Bitam, A. Mellouk, QoS Swarm bee routing protocol for vehicular ad hoc networks, in: IEEE International Conference on Communications (ICC’11), Kyoto, Japan, 2011. [9] S. Bitam, M. Batouche, E.-G. Talbi, A survey on bee colony algorithms, in: 24th IEEE International Parallel and Distributed Processing Symposium, NIDISC Workshop, Atlanta, GA, USA, 2010, pp. 1–8. [10] S. Bitam, M. Batouche, A. Mellouk, QoSBeeManet: a new QoS multipath routing protocol for mobile ad-hoc networks, 53rd IEEE Global Communications Conference (GLOBECOM’10), SACONAS Workshop, Miami, FL, USA, 2010, pp. 1648–1652. [11] Jeremy J. Blum, A. Eskandarian, L.J. Hoffman, Challenges of Intervehicle Ad Hoc Networks, in: IEEE Transactions on Intelligent Transportation Systems, vol. 5, no. 4, 2004, pp. 374–351. [12] G. Caizzone, W. Erangoli, P. Giacomazzi, G. Verticale, An enhanced GPSR routing algorithm for TDMA-based ad-hoc networks, in: 48th IEEE Global Communications Conference (GLOBECOM’05), St. Louis, MO, USA, 2005. [13] T. Camp, J. Boleng, V. Davies, A survey of mobility models for ad hoc network research, Wireless Communications and Mobile Computing 2 (5) (2002) 483– 502. [14] J. Chennikara-Varghese, W. Chen, O. Altintas, S. Cai, Survey of routing protocols for inter-vehicle communications, in: 3rd Annual International Conference on Mobile and Ubiquitous Systems (MOBIQUITOUS) Workshops, San Jose, CA, USA, 2006.

Please cite this article in press as: S. Bitam et al., HyBR: A Hybrid Bio-inspired Bee swarm Routing protocol for safety applications in Vehicular Ad hoc NETworks (VANETs), J. Syst. Architect. (2013), http://dx.doi.org/10.1016/j.sysarc.2013.04.004

S. Bitam et al. / Journal of Systems Architecture xxx (2013) xxx–xxx [15] T. Clausen, P. Jacquet, The Optimized Link-State Routing (OLSR) protocol, Internet Engineering Task Force Request for Comments 3626, 2003. [16] K.-T. Feng, C.-H. Hsu, T.-E. Lu, Velocity-assisted predictive mobility and location-aware routing protocols for mobile ad hoc networks, IEEE Transactions on Vehicular Technology 57 (1) (2008) 448–464. [17] S. Giordano, I. Stojmenovic, L. Blazevic, Position based routing algorithms for Ad Hoc networks: a taxonomy, in: X. Cheng, X. Huang, D.Z. Du (Eds.), Ad Hoc Wireless Networking, Kluwer Academic, Boston, Mass, USA, 2004, pp. 103– 136. [18] D.E. Goldberg, K. Deb, A comparative analysis of selection schemes used in genetic algorithms, Evolutionary Computation Journal, MIT Press 4 (4) (1996) 361–394. [19] J.H. Holland, Adaptation in Natural and Artificial Systems, University of Michigan Press, Ann Arbor, Michigan, MIT Press, 1992. [20] E. Hossain, G. Chow, V.C.M. Leung, R.D. McLeod, L. Misic, V.W.S. Wong, O. Yang, Vehicular telematics over heterogeneous wireless networks: a survey, Computer Communications, Elsevier 33 (7) (2010) 775–793. [21] D. Jiang, L. Delgrossi, IEEE 802.11p: towards an international standard for wireless access in vehicular environments, in: 67th IEEE Vehicular Technology Conference VTC-Spring, Marina Bay, Singapore, 2008, pp. 2036–2040. [22] D.B. Johnson, D.A. Maltz, J. Broch, DSR: The Dynamic Source Routing Protocol for Multi-Hop Wireless Ad hoc Networks, in: C. Perkins (Ed.), Ad hoc Networking, Addison-Wesley, 2001, pp. 139–172. [23] B. Karp, H.T. Kung, GPSR: greedy perimeter stateless routing for wireless networks, in: 6th IEEE/ACM annual International Conference on Mobile Computing and Networking (MobiCom’00), Boston, MA, USA, 2000, pp. 243– 254. [24] U. Korner, A. Hamidian, A. Nilsson, Performance of internet access solutions in mobile ad hoc networks, in: Wireless Systems and Mobility in Next Generation Internet, 1st International Workshop of the EURO-NGI Network of Excellence, Dagstuhl Castle, Germany, 2004. [25] K.C. Lee, U. Lee, M. Gerla, Survey of Routing Protocols in Vehicular Ad Hoc Networks, book chapter in Advances in Vehicular Ad-Hoc Networks: Developments and Challenges, IGI Global Edition, USA, 2010. [26] T. Li, S.K. Hazra, W. Seah, A position-based routing protocol for metropolitan bus networks, in: 61st IEEE VTC-Spring, Stockholm, Sweden, Jun. 2005, pp. 2315–2319. [27] B. Liang, Z. Haas, Predictive distance-based mobility management for PCS networks, in: 18th IEEE International Conference on Computer Communications (INFOCOM’99) Conference, NY, USA, 1999, pp. 1377–1384. [28] C. Lochert, H. Hartenstein, J. Tian, H. Füßler, D. Hermann, M. Mauve, A routing strategy for vehicular ad hoc networks in city environments, IEEE Intelligent Vehicles Symposium, Columbus, OH, 2003, pp. 156–161. [29] C. Lochert, M. Mauve, H. Füssler, H. Hartenstein, Geographic routing in city scenarios, SIGMOBILE Mobile Computing and Communications Review 9 (1) (2005) 69–72. [30] G. Nagib, W.G. Ali, Network routing protocol using genetic algorithms, International Journal of Electrical and Computer Sciences 10 (2) (2010) 40–44. [31] V. Namboodiri, L. Gao, Prediction-based routing for vehicular ad hoc networks, IEEE Transactions on Vehicular Technology 56 (4) (2007) 2332–2345. [32] V. Naumov, T.R. Gross, Connectivity-aware routing (CAR) in vehicular ad-hoc networks, in: 26th IEEE International Conference on Computer Communications (INFOCOM’07), Anchorage, Alaska, USA, 2007, pp. 1919– 1927. [33] The network simulator, ns-2, http://www.isi.edu/nsnam/ns. [34] V. Nundloll, G.S. Blair, P. Grace, A component-based approach for reconfigurable routing in VANETs, in: 8th International Workshop on Adaptive and Reflective Middleware, Urbana Champaign, USA, 2009. [35] J. Nzouonta, N. Rajgure, G. Wang, C. Borcea, VANET routing on city roads using real-time vehicular traffic information, IEEE Transactions on Vehicular Technology 58 (7) (2009) 3609–3626. [36] C. Perkins, E.M. Belding-Royer, S. Das, Ad Hoc On-Demand Distance Vector (AODV) Routing, Internet Engineering Task Force Request For Comments 3561, 2003. [37] K. Plößl, H, Federrath, A privacy aware and efficient security infrastructure for vehicular ad hoc networks, Computer Standards and Interfaces, vol. 30, no. 9, 2008, pp. 390–397. [38] M. Rondinone, J. Gozalvez, Distributed and Real Time Communications Road Connectivity Discovery through Vehicular Ad-hoc Networks, in: 13th IEEE International Conference on Intelligent Transportation Systems (ITSC), Madeira Island, Portugal, 2010, pp. 1079–1084. [39] H. Saleet, R. Langar, K. Naik, R. Boutaba, A. Nayak, N. Goel, Intersection-based Geographical Routing protocol for VANETs: a proposal and analysis, IEEE Transactions on Vehicular Technology 60 (9) (2011) 4560–4575. [40] J. Santa, M. Tsukada, T. Ernst, O. Mehani, A.F. Gómez-Skarmeta, Assessment of VANET multi-hop routing over an experimental platform, International Journal of Internet Protocol Technology 4 (3) (2009) 158–172. [41] S. Savasta, M. Pini, G. Marfia, Performance assessment of a commercial GPS receiver for networking applications, IEEE International Consumer Communications and Networking Conference, Las Vegas, NV, 2008, pp. 613– 617.

15

[42] K. Shafiee, V.C.M. Leung, Connectivity-aware minimum-delay geographic routing with vehicle tracking in VANETs, Computer Networks, Elsevier 9 (2) (2011) 131–141. [43] S.S. Wang, Y.S. Lin, PassCAR: a passive clustering aided routing protocol for vehicular ad hoc networks, Computer Communications 36 (2) (2013) 170–179. [44] S. Zeadally, R. Hunt, Y.S. Chen, A. Irwin, A. Hassan, Vehicular ad hoc networks (VANETS): Status, Results, and Challenges, Telecommunication Systems, vol. 50, no. 4, 2012. [45] J. Zhao, G. Cao, VADD: vehicle-assisted data delivery in vehicular ad hoc networks, IEEE Transactions on Vehicular Technology 57 (3) (2008) 1910– 1922.

Salim Bitam received the State Engineer degree in computer science from Mentouri University, Constantine, Algeria, in 1999 and the Magister and Doctorate in science degrees in computer science from Mohamed Khider University, Biskra, Algeria, in 2002 and 2011, respectively. In December 2002, he has been an assistant professor and since January 2011 he is an associate professor in computer science department and a senior member of Laboratory of Expert Systems, Imagery and their Applications in Engineering (LESIA) in University of Biskra. His main research interests are cloud computing, mobile ad hoc networks, vehicular ad hoc networks, wireless sensor networks and bio-inspired methods for optimization. Dr. Salim Bitam has to his credit more than 25 publications in journals, books and conferences, in which he has received two best paper awards. He has served as an editorial member and a reviewer of several journals such as Elsevier and Springer and on the technical program committees of several international conferences (IEEE Globecom 2012, WorldCIST 2013, etc.). He has participated at the Join-MED events in Egypt, Morocco and Tunisia sponsored by the European Union under grant agreement no. 231550.

Abdelhamid Mellouk is a full professor at University of Paris-Est (UPEC), Networks & Telecommunications (N&T) Department and LiSSi Laboratory, IUT C/V, France. He graduated in computer network engineering from the Computer Science High Eng. School, University Oran-EsSenia, Algeria, and the University of Paris Sud XI Orsay, received his Ph.D. in informatics from the same university, and a Doctorate of Sciences (Habilitation) diploma from UPEC. Founder of the Network Control Research activity with extensive international academic and industrial collaborations, his general area of research is in smart communications for adaptive realtime control for high-speed new generation dynamic wired/wireless networking in order to maintain acceptable quality of service/experience for added value services. He is an active member of the IEEE Communications Society and held several offices including leadership positions in IEEE Communications Society Technical Committees (Chair of the Technical Committee on Communications Software, Officer of The Technical Committee on Switching and Routing, etc.). He has published/coordinated several books and refereed international publications in journals, conferences, and books, in addition to numerous keynotes and plenary talks in flagship venues. He serves on the Editorial Boards or as Associate Editor for several journals, and he is chairing or has chaired (or co-chaired) some of the top international conferences and symposia.

Sherali Zeadally is an Associate Professor in the Department of Computer Science and Information Technology at the University of the District of Columbia, Washington, DC. His research interests focus on computer networks, including wired/wireless networks, network/system/ cyber security, mobile computing, ubiquitous computing, multimedia, and performance evaluation of systems and networks. Zeadally received his Doctorate degree in Computer Science from the University of Buckingham, England. He is a Fellow of the British Computer Society and a Fellow of the Institute of Engineering Technology, England. Contact him at [email protected].

Please cite this article in press as: S. Bitam et al., HyBR: A Hybrid Bio-inspired Bee swarm Routing protocol for safety applications in Vehicular Ad hoc NETworks (VANETs), J. Syst. Architect. (2013), http://dx.doi.org/10.1016/j.sysarc.2013.04.004