- Email: [email protected]
Efficient Dynamic-Power AODV Routing Protocol Based on Node Density
Journal Pre-proof
Efficient Dynamic-Power AODV Routing Protocol Based on Node Density Alwi M. Bamhdi PII: DOI: Reference:
S0920-5489(19)30445-3 https://doi.org/10.1016/j.csi.2019.103406 CSI 103406
To appear in:
Computer Standards & Interfaces
Received date: Revised date: Accepted date:
26 November 2019 18 December 2019 18 December 2019
Please cite this article as: Alwi M. Bamhdi , Efficient Dynamic-Power AODV Routing Protocol Based on Node Density, Computer Standards & Interfaces (2020), doi: https://doi.org/10.1016/j.csi.2019.103406
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier B.V.
Paper Title: Efficient Dynamic-Power AODV Routing Protocol Based on Node Density Alwi M. Bamhdi Umm Al-Qura University, College of Computers (AlQunfodah), Makkah, Saudi Arabia Email: [email protected] Corresponding Author: Alwi M. Bamhdi, [email protected]
ABSTRACT Mobile Ad Hoc Networks (MANET) allows versatile communication between host devices moving around in a state of flux. These networks have no fixed infrastructure thus making the routing of packets a continuous challenge to be optimally delivered under varying criteria such as when the number of nodes within an area increases the scope for interference between nodes which also increases significantly. Additionally, MANETs have low stability in areas with fast moving nodes which leads to their reduced longevity. This paper proposes a method by adapting the standard Ad hoc On-Demand Distance Vector (AODV) protocol to dynamically adjust transmission power usage, which is titled Dynamic Power-Ad hoc On-Demand Distance Vector (DP-AODV). This method uses the dependence of a transmission range on density to achieve this improvement. The results demonstrate that as density increases, DP-AODV shows decrease in delay than AODV and offering better performance for highly populated networks exceeding 200 nodes. The simulation results show that DP-AODV increase network throughput whilst reducing the node interference in a dense region, as well as it enhances the overall network performance with respect to the increased packet delivery fraction, reducing the control overheads and jitter, enhancing overall throughput, reducing interferences and finally, shortening end-to-end delay in medium to high density conditions. KEYWORDS: MANET, Interference, Density, Delay, Throughput, DP-AODV, AODV. 1. INTRODUCTION The last decade has seen a gigantic rise in the number of wireless network users due to the increasing availability of mobile devices with wireless capability [1,2,3,4]. Besides providing improved mobility for individual users, wireless networks have also allowed organizations to collaborate better, enhance their business flow and increase productivity. Wireless networks [5,6,7,8] are divided into two types: infrastructure and non-infrastructure. For infrastructure networks, base stations plays a key role in coordinating access to one or more transmission channels for mobile devices located within a coverage area and sharing the bandwidth of the wireless channels. GSM and Wi-Fi are the typical examples of such networks. Non-infrastructure networks have no fixed base station or wired getaway and is formed by a collection of mobile nodes on an ad hoc basis. MANETS, in the form of Ad hoc networks and sensor networks are examples of non-infrastructure networks. Ad hoc networks [9, 10] are also known as temporary networks and are used for applications where quick deployment and sharing of information are important. These networks mainly consist of sets of independent mobile nodes that dynamically change their positions so as to automatically establish and maintain routes from sources to destinations. The mobile nodes act both as end systems and also as routers. Crisis management services for disaster recovery and military applications are examples of applications requiring ad hoc networks. The rest of the paper is organized as follows: Section 2 provides an overview of routing in ad hoc, AODV a, Section 3 describes the DP-AODV proposed method, Section 4 discusses the transmission power model, Section 5 describes the simulation environment and the performance metrics, the performance results, and the observation. Finally, the conclusion is provided in Section 6.
1
2. LITERATURE REVIEW In general, the primary purpose of a network routing protocol is to allow packets to be efficiently directed and delivered to their ultimate destinations. Routing protocols for mobile ad-hoc networks have different requirements compared to that of wired networks, and are mainly oriented towards the reduction of the number of transmissions to shorten end-to-end delay optimally rather than any other metric such as throughput. Routing protocols in these types of networks are often classified as either proactive or reactive. Proactive protocols attempt to maintain routes to all nodes in the network, whereas, reactive protocols only generate routes when they are required for transferring data at that instant.
2.1 Ad hoc On-Demand Distance Vector Protocol The AODV [13, 14] algorithm is a reactive routing protocol based on DSDV routing protocol. AODV is hop-byhop in which each intermediate node decides where the routed packet must be forwarded next. As a result, the node in routing contains a route table to maintain new route information that has three important fields: a hop count, next hope node and a sequence number. It introduces low overhead, memory overhead, quick adaptation to dynamic link conditions and low processing. AODV is dependent on the distance vector algorithm. It requests a path when necessary and does not require nodes which are not actively used in connection to maintain routes to the destination. There are some features of AODV, including loop release and link breakages, which lead to the property group of nodes being notified. The algorithm uses individual packets to find and maintain links. Another benefit is the hello message which is broadcast at periodic intervals to the immediate neighbours. AODV has a multicast route invalidation message. AODV routing protocol consists of three different message types: Route Request (RREQ), Route Reply (RREP) and Route Error (RERR). When a node wants to send a data packet (message) to a destination, firstly, it will examine the route table to decide if the route to the endpoint of communication is a valid connection or not. If the route is valid, this node will decide to send a message to the next hop node. If not, it begins route discovery by broadcasting RREQ packet. A RREQ packet consists of a broadcast ID, a source node IP address, the last known sequence number of the destination and a new sequence number for the source. Each node that receives the RREQ should update information for the source node and establish backward pointers to the source node in the route table. The RREQ arrives either at the destination or at an intermediate node from where there is a route to the destination where the RREP is made. Once the real route is made by unicasting a RREP to the origin of the RREQ, every node receiving the RREQ caches a route back to the origin of the RREQ. The intermediate nodes send back RREP to the previous node and become a part of the route to the destination. If the nodes receive the same RREQ packets later, they ignore it and do not send it on. In the case of the source node, it refreshes the entry in the route table and uses this route in the future. Routing of packets is one of the most important challenges facing MANETs due to unstable links, dynamic features and limited resources of these networks [11]. Several attributes of MANETs contribute to the complexity of designing ad hoc routing protocols: density, mobility, device heterogeneity, quality of Service (QoS), battery constraints and bandwidths. One of the main challenges facing mobile communication is transmitting a packet at fixed transmission power that may drain the power consumption of the communication interface. Packets transmitted at a common maximum power trigger a considerable loss on power consumption. Consequently, when node pairs are close to each other, then the power transmission required for them to communicate may be kept at a minimum. In such a scenario, transmitting packets at high power level may generate significant interference to the network and consume more power than required. Thus, an appropriate transmission power for each packet needs to be considered, referred to as a transmission power control. This requires a power level either high enough to guarantee the transmission or low enough to provide less interference and high throughput. Power control minimizes the transmission power by controlling the transmitting range. A dynamic transmission range is more effective in maintaining connectivity while minimizing adverse effects of a high transmission power. Worth noting that power control in ad hoc networks is more important than in cellular networks as the nodes in an ad hoc network 2
communicate by sending data to neighboring nodes. Node density has a great impact on the performance of ad hoc networks in that it influences factors such as routing efficiency, capacity and delay. In 1978, Silvester and Kleinrock published their paper “Optimum Transmission Radii for Packet Radio Networks” [12]. In this paper, the authors explore the trade-off between increased communication radius due to fewer hops used to reach the destination and, the effective channel throughput loss at each node as a result of the increase in transmission range. The paper proves that the best number of neighbours for a given node is 5.89 (rounded to six), and concludes that a node's transmission radius must be adjusted so that it has six neighbours. For a stationary network, the optimum connectivity is seven or eight neighbours per node, which is similar to Kleinrock's conclusion of six neighbours per node [13]. A power control scheme based on the number of neighbours in mobile ad hoc networks was proposed in [14]. In that study, the authors investigated both physical and connectivity design issues. In [15], a mechanism to reduce the transmission and reception power for the frequently used nodes was presented, while another study[16] looked into the transmission power optimization algorithm based on various nearest-neighbor distances algorithm. In [17], the authors proposed an adaptive transmission power control scheme based on the autonomous clustering. In their proposed scheme an estimate of the distance between the current and neighboring node numbers is used to control the transmission power at each node. The work presented in [18] investigates the power management in ad hoc networks by determining the effects of transmission power on network throughput and power consumption. The authors in [18] either allow or disallow nodes to dynamically adjust their power levels per transmission to be the maximum amount of power needed to reach the destination node. In another study [19], the authors present algorithms which adaptively adjust the transmission power of the nodes in response to topological changes. The goal of that study is to maintain a connected network while using minimum power. This paper examines the effect of using different transmission power levels for different number of neighbours by dynamically adjusting the transmission power to minimize the power and to reduce the interference between nodes to achieve a better throughput in the network. AODV protocol has two main characteristics which allow for transmission power improvement: 1)
It relies on dynamically establishing route table entries at intermediate nodes instead of source routing (e.g. DSR) and,
2)
It combines the behavior of DSDV and DSR routing protocols that perform generally well in most cases and are very effective in larger networks.
3. PROPOSED METHOD 3.1 Dynamic Power-AODV Routing Protocol In this paper, we propose a Dynamic Power Ad hoc On- Demand Distance Vector (DP-AODV) protocol which is an improvement of the existing AODV routing protocol. In this extension, we modify all the packets, so that, every packet dispatched from the routing layer will contain the modified packet header with the distance information of the destination and neighbours count. The transmission power is only decided in a wireless physical layer just before transmitting the packet and calculating the transmission power at the wireless physical layer by using the distance information. The hello packets of AODV are modified to carry the “x-y position” coordinate field information. This is done in order to obtain the exact location information of a node which was used by the routing protocol to determine the route. When a node receives the hello message, it calculates the distance for neighbours using the embedded coordinates. Distance of the neighboring nodes is an important aspect in route discovery since all the nodes within the coverage of a particular node can receive the route request messages and process them. Consequently, increasing or decreasing the transmission power of a node increases or decreases respectively the number of one hop neighbours involved in the surrounding domain neighborhood. So, our algorithm is selecting the power needed to maintain the connectivity among the nodes and hence reduce the overall power. Then, each node is added to the neighboring table with the transmission power and distance. Along with the normal information which will be in the AODV routing table, we are maintaining the distances for all the neighboring nodes in the routing table and the routing table entry has not being changed. Every packet is leaving from a node that will contain the distance information about the destination node. So, at the wireless physical layer, while sending the packet to 3
that neighbor node, the transmission power will be changed according to the “level” of power needed to reach the destination node. Fig. 1 shows three scenarios of low, medium to high transmission density in typical dynamic power ad hoc network.
Fig. 1. A Dynamic Power Ad Hoc Network This implies that the selected path will be a stable one based on the densities of the different locations and consequently may not necessarily be the shortest path. The power changing is done in the physical layer and it has a power file to store the transmission power and values in the file. If a packet is scheduled for transmission, it will reach the physical layer (the routing agent will add the distance and the neighbor count inside the packet). While reaching the physical layer, the packet contains the node's neighbor count and the distance. The transmission power can be changed with respect to the neighbor’s count, the distance and the transmitter. DPAODV uses different power levels to determine a route for transmission of a packet. For a small number of neighbours, transmission is done at maximum power; otherwise transmission power is reduced accordingly Fig 1. As our proposal is based on an AODV routing protocol, the fundamental mechanism of data packet handling is the same. However, in our protocol, the basic mechanism of a neighbor based Variable-Power Transmission includes seven steps: 1.
Initially, where each node broadcasts a hello message with its coordinates to determine its neighbours.
Each node uses broadcasting mechanism which sends a packet to every neighbor and the hello packets are generally broadcasted only up to one hop neighbours. But, normally, AODV uses network wide multi-hop broadcast. If a node has already a reduced power to maintain the required number of neighbours, then, the one hop broadcast only reaches up to that distance with respect to that node’s current transmission power. In the present implementation, all the updates in the distance information on the routing table is handled using existing mechanisms of AODV to avoid unnecessary increases in overheads. Each node position is updated dynamically using hello message as shown in Fig 2 with “coordinates” information added as a parameter. Type
Reserve
Destination IP Address Destination Sequence Number Hop Count Lifetime Coordinates Fig. 2. Hello Packet Format in DP-AODV
4
2.
The neighbour node receives a hello message and calculates the distance and adds the entry in the neighbor’s table.
Each node knows its neighbor’s information and the neighbor table contains the neighbour’s_Id, txpower, distance, flag and the neighbor’s count that is used to maintain the neighborhood information. In the Euclidean distance formula, If x = (x1, x2) and y = (y1,y2) then the distance is given by: d (x , y)= √(x2 – x1)2 + (y2 – y1)2 The distance is estimated using Euclidean distance which is only based on the certain coordinates of the nodes that are received through the hello packet. The nodes are sending and updating its location through a hello reply message. When the node knows the distance then that would be the best transmission power level to choose. 3.
Identify the power level for the different number of neighbours.
In our implementation, while transmitting a packet to a destination node, the transmitting node will select the power level based on the number of neighbours and the power required for reaching the destination. 4.
During data transmission the node estimates a number of neighbours by using the neighbour’s count function.
The neighbor count is already available in default AODV through a hello request and reply mechanism. We just add the distance of the neighbor using the same hello mechanism and using the number of entries found in the neighbor’s table.
5.
Transmission power required based on the power level with respect to the number of neighbours.
Collisions within the interference range of a domain/plane are avoided by reducing the transmission range of active neighboring nodes along the route such that they converge to a minimum distance needed to maintain connectivity. The neighboring scheme comprises of four main elements: • • • •
Measurement of the estimated critical range within the domain; Estimation of an ideal node’s power requirement; Allowing a tolerance factor percentage for nodal mobility, some transmission noise and interferences; Selection of the ideal power levels for that node.
When periodic beacons such as the hello packets are received, the node calculates the critical transmission range in the density areas. This is then used to estimate and select the transmission power so that the performance of the routing protocol can be enhanced appropriately. After obtaining the calculated ideal power value, it is periodically checked to ensure that it remains within the designated transmitted power level (Pt) by the receiver. This is to avoid nodes that are currently not involved in the sending of any data packets from adjusting their transmission powers dynamically to a value that is outside the lower and upper bounds to eliminate wastage of energy resources by the participating nodes.
Level 1 If density is low, < 7 neighbours Pt =
Level 2 If density is medium, between 7 and 15 neighbours Level 3 If density is high, > 15 neighbours
5
As an example, in our experiment, we used three different transmission power levels, depending on different number of neighbours involved in the density area range. When the number of nodes is less than 7 neighbours at the source, then Pt = Level 1 in the range which transmits at full power such as 250m. If the number of neighbours is higher than 15 neighbours, then Pt = Level 3 in the range which transmits at a reduced power such as 170m. When a node can be reached only by using the power needed to transmit up to the cross-over distance, then using any power more than that is not necessary. So, at that location, the node will use only a lower level of power to maintain connectivity. Similarly, we use different levels of powers to maintain connectivity at different density areas. 6.
Adjusting the transmission power
Based on the estimated distance transmission power required to transmit the packet to a corresponding neighbor depends on two ray ground propagation. The transmission power is required to reach the next hop or destination. Our algorithm consists of an input parameter, such, as the variable-power. This parameter can adjust the power level based transmission depending on the different number of neighbours. 7.
Forward the packet
It is to transmit the packet using minimum energy required. Each packet handled in the wireless physical layer of the current node will be transmitted as per the number of neighbours of the nodes as well as the next hop is resolved from the routing table (destination). 4. TRANSMISSION POWER MODEL The free space propagation model [23] and the two-ray propagation model [24] are used to calculate the received power of each packet between the neighboring nodes. The free space assumes that the ideal propagation condition of a clear line-of-sight path between the transmitter and receiver exist, and the two-ray propagation assumes that the ground refection path besides the line-of-sight propagation is observed. The relation between the transmitted signal power (Pt) and the received signal power (Pr) is given as:
𝑃𝑟 =
𝑃𝑡 ∗ 𝐺𝑡 ∗ 𝐺𝑟 ∗ 𝜆2 ]1[ (4𝜋)2𝑑2𝐿
Where, Pr : Received Signal Power, Pt : Transmitted Signal Power, Gt and Gr are Transmission and Receiver Gain of Antenna (1.0), d : Distance between the Transmitter and Receiver, dc : Cross Over Distance , L : System Loss, λ : Wavelength. In addition to the line-of-sight path, the two-ray model also takes account of the ground reflection of the path. This model gives a more accurate prediction for longer distances, where hr and ht are Height of antenna for receiver and transmitter (1.5 m). The equation for this model is:
𝑃𝑟 =
𝑃𝑡 ∗ 𝐺𝑡 ∗ 𝐺𝑟 ∗ℎ𝑡2∗ℎ𝑟2 ]2[ 𝑑 4𝐿
If the distance is less than, or equal to, the cross-over distance, then the free-space propagation model is used.
6
Otherwise, the transmission is regarded as a two-ray propagation model. The equation used to calculate the crossover distance is as follows:
𝑑𝑐 =
4𝜋∗ℎ𝑡 ∗ℎ𝑟 𝜆
]3[
Where, dc: Cross Over Distance, which is the reference to the distance of the receiver. Transmission between a transmitting node and a receiving node is only successful if the received power of the radio signal is above a certain threshold. However, if the received power is below this threshold then the transmission is undetected. Moreover, if there are multiple transmissions at the same time, then only the transmission with the highest power is received. The standard NS-2 radio propagation models comprising of the free space propagation model and the two-ray propagation model are used in our simulation. 5. EXPERIMENTAL EVALUATION In our study, we used a Network Simulator 2 (NS-2) [22], which is very popular in the data communication and computer networking technological, research and development community. It is an event-driven simulation tool that is used in order to analyze and evaluate the DP-AODV communication protocol. It supports protocols including AODV, DSDV, DSR, AOMDV and TORA and consists of two main programming languages: C++ and OTcl. 5.1 The Simulation Environment Other than defining the requirements and design of DP-AODV, the methodology and experimentation for implementing the DP-AODV protocol is simulation-based prototyping. It extends the existing AODV protocol by adding some fields in the NS-2 packet header and uses the NS-2 simulator to test and evaluate the enhanced protocol in a variety of scenarios, and to compare the results with AODV routing protocol. Simulation is used to test the protocol as, unlike real-life experiments, it is easy to conduct, less expensive and enables the adjustment of parameters for different scenarios. A Tcl scripting file is an input into the tool, an ASCII trace file corresponding to the event registered at network level and organized according to certain fields and a network animator visualization tool (called NAM) are used for displaying the nodes in the network in real-time, which are also very useful to verify the accuracy of the simulation. NS-2 primarily outputs text-based simulation results. The diagram in Fig. 3 represents the modified modules. C++ is used under NS2 in order to implement the DP-AODV routing protocol, with the use of TCL (Tool Command Language) scripts to describe the simulation scenarios. It shows where our extensions are arranged within the NS2 framework.
7
Fig. 3: Files in the NS-2 Simulator Framework that were modified. The files for the modified modules are as follows: Mac/wireless-phy.cc: NS-2 contains an energy model for wireless nodes that is useful for assessing the advantages of different energy conservation techniques, such as the sleep mode, or using the optimal network densities. The model allows the power requirements for transmitting and receiving the packets, or for an idle mode, to be specified. In Mac/wireless-phy.cc, we have included the routing agent that will add the distance of the destination and the neighbor count inside the packet. Mac/wireless-phy.h: This is the header file where we defined the power timer and power file (double power, double time) to store all the events. Aodv/aodv.cc: It contains, sends and receives hello messages, sends and receives aodv function and the routing table to maintain the information of each node. In this file, all timers, routing agents and Tcl hooks are actually implemented. Aodv/aodv.h: This header file is where we define the mobile node, getDistance (double x, double y), number of neighbours, flag and neighbor tables. We also define the neighbours list (neighbor IDs, txpower and distance). Aodv/packet.h: This is the header file where the AODV reply header (xpos, ypos and transmission power) is declared. Aodv/rtable.h: This is the header file where the routing table and neighbor’s table (txpower and distance) are declared. Common/packet.h: Each packet in NS-2 is associated with a unique type that associates it with the protocol to which it belongs, such as TCP , ARP , AODV , FTP , etc. We defined x, y, txpower, num of nbrs in the packet.h header file. Common/packet.cc: This file contains the size of a packet’s header, free list, off- set of common header and offset of flag’s header, which is accessible through Tcl. It is used to manage active packet header types. Each packet in NS-2 is used to exchange information between objects in the simulation. We defined num of nbrs in the packet.cc source file. The structure of the protocol was created using an agent, which represents the endpoints and can be used to implement the protocol at various layers. The agent is the principal class for implementing the protocol and provides
8
a link with the Tcl interface, for control of the protocol using TCL scripts. New control packet parameters are defined by DP-AODV, in the common packet.h header file, to represent the format of the control packets. The protocol can send packets periodically or following a delay after the occurrence of an event. We also used random scenarios that have different parameters each of which have an effect on the ad hoc movement pattern. The parameters which can be altered are: Maximum speed: a speed is assigned randomly between (0-30m/s) every time. Topology: different number of nodes (75,100,150, 200).
A dynamic network is when the nodes move continuously throughout the simulation period, which is the worst case scenario for the network performance. A static network has very low mobility where nodes are completely static. Initial energy for each node is set at 100 Joules at the start of every simulation in each scenario, and a node will consume 0.9W for sending packets and will consume 0.6W for receiving packets. The maximum speed of the nodes was limited to 30 m/s. The traffic is generated by 40 CBR sources distributing the traffic among all nodes. The sending rate was fixed at 4 packets per second, with a data packet size of 512 bytes. The proposed DP-AODV model assumes random distribution of nodes within a plane or domain in a typical ad hoc network. We used a fixed area in the simulation to ensure consistency and subsequent comparability to the standard AODV protocol. The simulation parameters which are used for the node movement scenario are shown in Table I. Table I. Movement Node Parameters Parameter Value NS-2.34 Simulator version Random waypoint Mobility model IEEE 802.11 Physical/MAC layer DP-AODV and Protocols AODV Free space & TwoPropagation ray Models Omni-Antenna Antenna Model Random Node-placement 2 (Mbps) Bandwidth 1000×1000 (m) Simulation area 100 (sec) Simulation time 75,100,150 and 200 Node Density 0~30 (m/s) Speed 40 (sec) Pause time 10 Iteration Traffic is generated using a constant bit rate from the source nodes. The traffic pattern numbers used by the source nodes are listed in Table II.
9
Table II. Traffic Pattern Parameters Parameter Traffic type Packet size Max connection Max sending packet Initial energy setting Txpower of nodes Rxpower of nodes
Value CBR (UDP) 512 40 4 packets/sec (2 Mb/sec) 100 Joules per node 0.9 (W) 0.6 (W)
5.2 PERFORMANCE METRICS In order to evaluate the performance, several metrics can be used, however for the purposes of this research the following metrics were given for routing protocols evaluation: Packet Delivery Fraction (PDF)
The ratio of received packets by CBR, sink at destination over sent packets by the constant bit rate source (CBR, application layer). The metric tells us how reliable the Protocol is. End-to-End Delay
The cumulative delay that might come about as a result of buffering during discovery of routes queuing at interfaces, delays in retransmission at the MAC, and the time taken for propagation and transfer. Throughput
The amount of data that is transmitted per unit time, (i.e., data bytes delivered to their destination per second. Control Overhead
The ratio of the total number of control packets to the total number of packets sent to the MAC layer. Jitter
Measures the variation in transmission time of packets received at the destination, caused mainly by network congestion, delivery time delays, timing drift or route changes to reach the destination. This mainly happens when a system is in a nondeterministic state. Worth mentioning, the most important metrics for best-effort traffic are the packet delivery fraction and the endto-end delay. However, these metrics are not entirely independent as a shorter delay does not necessarily mean a higher packet delivery fraction, as only successfully delivered packets are used to measure the delay. A lower packet delivery fraction and longer delay may, however, cause a larger overhead. 5.3 PERFORMANCE RESULTS The results of the analysis carried out offer a clear demonstration of the advantages of the improvements made in DP-AODV in comparison with basic AODV. The comparison of the results from the analysis is more realistic and accurate as both the routing protocols used single-path transmission in the experimental scenarios. Varying movement patterns and traffic patterns scenario files were run with both routing protocols in order to ensure a fair comparison without bias. 10
From Fig 4, it can be seen that the packet delivery fraction has always improved in different kinds of densities with the use of DP-AODV compared to standard AODV. It has improved from 12 % to 31% because the proposed method restricts the density by controlling the transmission power. The dropped packet is high in basic AODV than DP-AODV, and many packets are lost due to fixed transmission power and it makes an extensive interference between nodes. As can be seen from Fig 5, the delay has reduced in most of the densities with DP-AODV. It has decreased to 52%. Hence, for routes having a higher success probability and that use less time to transmit data, DPAODV is generally less than that of basic AODV. These results occur as the proposed DP-AODV restricts the density by controlling the transmission power, but with AODV many packets are lost due to fixed transmission power and extensive interference between nodes.
PAKET DELIVERY RATIO (%)
30 m/s Speed 100 80 60 40 20 0 75
100
150
200
DENSITY DP-AODV
AODV
Fig. 4. Packet Delivery Fraction (PDF) for various node densities
30 m/s Speed DELAY (S)
2 1.5 1 0.5 0 75
100
150
200
DENSITY DP-AODV
AODV
Fig. 5. Delay for various node densities Fig 6 shows that DP- AODV gives a shorter EED as compared to AODV because in a dynamic network as the density increases, the EED for DP-AODV increases, though there is a decrease from 150-200 nodes. Likewise, with AODV, as the density increases, the EED increases. Similarly, in a static network, as the density increases, the EED for DP-AODV increases. Likewise, with AODV, as the density increases, the EED increases. The reason for this is that finding a routing with a higher success possibility (as observed for DP-AODV) will definitely use less time to send data from the source node to the destination node.
11
Fig. 6: EED vs Density (Pause Time 0 sec - Dynamic) It can be seen in Fig 7, the throughput of the networks between DP-AODV and basic AODV increases the density. The throughput of the network in DP-AODV is much more enhanced than the basic AODV. The reason is due to the reduction of bandwidth waste by RREQ data packet in the route discovery. From Fig 8, it may be seen that the overhead of the DP- AODV is less than the basic AODV because it sends out less number of control packets that can reduce the overhead in DP- AODV. AODV has a high overhead rate due to use of a large share of network resources. The Fig 5 shows that the delays between packets decrease in DP-AODV and perform well despite node density, while a basic AODV performs poorly in terms of link breakage and jitter conditions as seen in Fig 9. The density performance improves as nodes density within the region increases. When we reduce the transmission level with respect to the distance of the next hop to the final destination then there is no possibility of frequent link failure. For an optimum level of power usage during multi-hop forwarding of packets, each region attempts to use minimum power to decrease the overall communication overhead and invariably reduce power consumption. Generally, DPAODV exhibits better performance and is more efficient than AODV in high density networks as node density gradually increases. In low density, AODV performs better but there is high probability for some errors to occur during the routing discovery process when using the default transmission power. In a dense AODV network, there may be more collisions between neighboring nodes, which means that more energy is wasted at the constant radio level, which is avoided in DP-AODV.
THROUGHPUT (KBIT/S)
30 m/s Speed 250000 200000 150000 100000 50000 0 75
100
150
200
DENSITY DP-AODV
AODV
Fig. 7. Throughput for various node densities
12
CONTROL OVERHEAD (%)
30 m/s Speed 150000 100000 50000 0 75
100
150
200
DENSITY DP-AODV
AODV
Fig. 8. Overhead for various node densities
30 m/s Speed JITTER (SEC)
0.15 0.1 0.05 0 75
100
150
200
DENSITY DP-AODV
AODV
Fig. 9. Jitter for various node densities In summary, the results have shown that DP-AODV performed better than basic AODV across all five of the metrics analyzed. DP-AODV gave a higher packet delivery fraction rate than AODV, particularly in a dynamic environment where the speed of nodes and number of sources where varied. DP- AODV also gave a shorter end-to-end delay than AODV across all metrics, with a typically lower end-to-end delay in a static environment, though there were some minor differences in dynamic environment experiments. The throughput results were also higher for DPAODV, particularly in a static environment when varying the speed of nodes and number of sources. The power consumption results revealed that DP-AODV has lower power consumption than basic AODV, and that power consumption for both protocols was lower in a static environment, with the exception of DP-AODV when the number of sources was varied. The power consumption for DP-AODV remained constant when the pause time was varied. With the final concluding remark that the proposed DP-AODV enhances the overall network performance with respect to increased packet delivery fraction, reduced control overheads and jitter, enhanced overall throughput, reduced interferences and finally, shortened end-to-end delay in medium to high-density conditions, which has significant impact in ad hoc networks. 5.4 FURTHER OBSERVATION Density is a key aspect of ad hoc networks, since high density can cause collision and congestion due to fixed transmission power which leads to high loss of packets in the networks. Another key aspect is the mobility. Since high mobility may cause frequent direct line breaks which could lead to unnecessary re-route discovery, hence waste of a lot of power. The expected outcomes of dynamically changing the transmission power scenario typically means that 13
a node initially increases its transmission power until it detects a set of neighbours in its vicinity and adjusts its power according to the number of neighbours. As the transmission power increases, the number of neighbours increases, and as the transmission power decreases, the number of neighbours decreases, enhancing the network throughput. Ultimately, the transmission power varies between minimum and maximum values. Although there has been considerable research in the field of power varying protocols, the majority of this type of power management protocol, for example COMPOW [25], aims to conserve power and do not specifically address the idea of dynamically reducing interference in the wireless network. However, the enhancements made to the existing AODV protocol by the DPAODV are intended to reduce mutual interference of radios as this directly impacts on energy consumption within the network. The dynamic variation of transmission power usage in order to reduce interference has recently gained in importance within the field of power management research as 3rd Generation Partnership Project (3GPP) standards organization which develops protocols for mobile telephony and networks has defined "5G NR" (5G New Radio) software as "5G". As mentioned before, the results demonstrate that the enhanced protocol (DP-AODV) generates better performance results as compared with a basic AODV. DP-AODV has different characteristics to AODV as it has a higher packet delivery fraction and throughput, a shorter end-to-end delay, lower power consumption and minimum jitter. These key characteristics are the results of both the power control mechanism and the enhanced hello message exchange mechanism that provide both a better success rate for packet delivery and demonstrate that the enhanced algorithm is more energy efficient, which is a major issue for ad hoc networks. The improved performance of DP-AODV compared to AODV can be attributed to several design factors. One of the major factors is the incorporation of the enhanced hello mechanism, which lowers the rate of problems in the algorithm to search for neighboring nodes. Nodes use hello messages to dynamically update the routing information which ensures a more stable link, without unnecessary link breakages, and as a result more successful packet delivery to destination nodes. 6. CONCLUSION The current paper proposed a technique that considered a number of neighbours and a level of power and transmission mechanism, which solved several problems in the network by improving an exciting AODV routing protocol. Our proposed technique evolved around the power level which dynamically increased or decreased, based on node density that was far more effective than fixed transmission power on network throughput. The proposed algorithm automatically adjusted the transmission power at each node so as to keep its number of neighbours within a specified range. Consequently, this lead to the prevention of unwanted interference and unnecessary overhearing and overprocessing by other nodes and thus increased the overall throughput in the network. The simulation results showed that DP-AODV enhanced protocol offered better network performance; hence improved the packet delivery, throughput and reduced interference as compared to AODV routing protocol. In our future work, we will attempt to improve the performance of DP-AODV routing protocol, by finding the appropriate threshold, the appropriate power level to use and investigate the consumed power involving a larger number of wireless nodes and more types of ad hoc network topologies to find a new metrics and measurement that will combine the delivery ratio and the energy consumption. We also expect to decrease the overhead, interference between nodes, and improve the throughput in the network to be adapted in high density and high mobility networks. In addition, we also will attempt to implement DP-AODV future wireless networks 5G cognitive radio networks and IEEE 802.16 based mesh networks in the context of Internet of Things (IoT).
14
Declaration of interests The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
7. REFERENCES [1] Munisha Devi and Nasib Singh Gill. Mobile Ad Hoc Networks and Routing Protocols in IoT Enabled Smart Environment: A Review. Journal of Engineering and Applied Sciences, 14(2019) 802-811. [2] Guaya-Delgado, L., Pallarès-Segarra, E., Mezher, A.M. et al. A novel dynamic reputation-based source routing protocol for mobile ad hoc networks. J Wireless Com Network 2019, 77 (2019) doi:10.1186/s13638-019-13757 [3] Jhaveri, Rutvij H., et al. "Sensitivity analysis of an attack-pattern discovery based trusted routing scheme for mobile ad-hoc networks in industrial IoT." IEEE Access 6 (2018) 20085-20103. [4] S. L. Meshram and P. D. Dorge, "Design and performance analysis of mobile Ad hoc network with reactive routing protocols," 2017 International Conference on Communication and Signal Processing (ICCSP), Chennai, (2017) 0443-0447.doi: 10.1109/ICCSP.2017.8286396 [5] Ejmaa, A. M. E., Subramaniam, S., Zukarnain, Z. A., & Hanapi, Z. M. Neighbor-based dynamic connectivity factor routing protocol for mobile ad hoc network. IEEE Access, 4 (2016) 8053-8064. [6] Han, Nguyen Dinh, Yonghwa Chung, and Minho Jo. "Green data centers for cloud-assisted mobile ad hoc networks in 5G." IEEE Network 29 (2015) 70-76. [7] H. Xiao, D. M. Ibrahim and B. Christianson, "Energy consumption in Mobile Ad Hoc Networks," 2014 IEEE Wireless Communications and Networking Conference (WCNC), Istanbul, (2014) 2599-2604. doi: 10.1109/WCNC.2014.6952818 [8] CR Komala, P.S. Srinivas Shetty, and E. Elevarasi, “Wireless Ad Hoc Mobile Networks,” In National Conference on Computing, Communication and Technology, pp. 168–174, 2010. [9] Mobile Ad hoc Networks (MANET). (2012). [Online]. Available: http://datatracker.ietf.org/wg/manet/charter/, IETF Working Group. [10] A.H.A. Rahman and Z.A. Zukarnain. “Performance Comparison of AODV, DSDV and I-DSDV Routing Protocols in Mobile Ad Hoc Networks,” In European Journal of Scientific Research, vol. 31, no. 4, pp. 566– 576, 2009. [11] A. Boukerche, B. Turgut, N. Aydin, M.Z. Ahmad, L. Boloni, and D, “Turgut. Survey paper: Routing Protocols in Ad Hoc Networks: A Survey Computer Networks,”: In the International Journal of Computer and Telecommunications Networking, vol. 55, no. 13, pp. 3032–3080, 2011. [12] L. Kleinrock and J. Silvester, “Optimum Transmission Radii for Packet Radio Networks or Why Six is a Magic Number,” In Proceedings of the IEEE National Telecommunications Conference, vol. 4, pp. 1– 4, Birimingham, Alabama, 1978. [13] E.M. Royer, P.M. Melliar-Smith, and L.E. Moser, “An Analysis of the Optimum Node Density for Ad Hoc Mobile Networks,” In International Conference on Communications, vol. 3, pp. 857–861, IEEE, 2001. [14] B. Abasgholi, R. Kazemi, M. Arezoomand, and A.R. Enayati, “Neighbor Detection Power Control (NDPC) MAC Protocol in Mobile Ad Hoc Networks,” In 3rd International Symposium on Wireless Pervasive Computing., pp. 508–512, IEEE, 2008. [15] H. Abusaimeh and S.H. Yang, “Reducing the Transmission and Reception Powers in the AODV,” In International Conference on Networking, Sensing and Control, pp. 60–65, IEEE, 2009. [16] Y. Chen, H. Yang, B. Liu, and J. Cheng, “Transmission power optimization algorithm in wireless ad hoc networks,” In International Conference on Communications and Mobile Computing., vol. 3, pp. 358- 363, IEEE, 2010. [17] R. Oda, T. Ohta, and Y. Kakuda, “A transmission power control adaptive to variation of node density in mobile ad hoc network clustering,” In Second International Conference on Communication Software and Networks, 15
pp. 141–146, IEEE, 2010. [18] T.A. ElBatt, S.V. Krishnamurthy, D. Connors, and S. Dao, “Power Management for Throughput Enhancement in Wireless Ad-hoc Networks,” In IEEE International Conference on Communications., vol. 3, pp. 1506–1513, IEEE, 2000. [19] R. Ramanathan and R. Rosales-Hain, “Topology Control of Multihop Wireless Networks using Transmit Power Adjustment,” In Nineteenth Annual Joint Conference of the IEEE Computer and Communications Societies, Proceedings., vol. 2, pp. 404–413, IEEE, 2000. [20] I.D. Chakeres and E.M. Belding-Royer, “AODV Routing Protocol Implementation Design,” In 24th International Conference on Distributed Computing Systems Workshops. Proc., pp. 698–703, IEEE, 2004. [21] C.E. Perkins, E. Royer, and S. Das, “Ad Hoc on Demand Distance Vector (AODV),” Request for Comments (RFC), 2003. [22] The Network Simulator ns-2. [Online]; Available: http : // www.isi.edu/nsnam/ns/. [23] C. S. R. Murthy and B. Manoj, Ad hoc wireless networks: Architectures and protocols. Prentice Hall, 2004. [24] C. E. Perkins, Ad hoc networking, vol. 1. Addison-wesley Reading, 2001. [25] S. Narayanaswamy, V Kawadia, R. S. Sreenivas and P. R. Kumar, “The COMPOW protocol for power control in ad hoc networks: Theory, architecture, algorithm, implementation, and experimentation,” In European Wireless Conference, pp. 156-162, 2002.
16
Efficient Dynamic-Power AODV Routing Protocol Based on Node Density Alwi M. Bamhdi PII: DOI: Reference:
S0920-5489(19)30445-3 https://doi.org/10.1016/j.csi.2019.103406 CSI 103406
To appear in:
Computer Standards & Interfaces
Received date: Revised date: Accepted date:
26 November 2019 18 December 2019 18 December 2019
Please cite this article as: Alwi M. Bamhdi , Efficient Dynamic-Power AODV Routing Protocol Based on Node Density, Computer Standards & Interfaces (2020), doi: https://doi.org/10.1016/j.csi.2019.103406
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier B.V.
Paper Title: Efficient Dynamic-Power AODV Routing Protocol Based on Node Density Alwi M. Bamhdi Umm Al-Qura University, College of Computers (AlQunfodah), Makkah, Saudi Arabia Email: [email protected] Corresponding Author: Alwi M. Bamhdi, [email protected]
ABSTRACT Mobile Ad Hoc Networks (MANET) allows versatile communication between host devices moving around in a state of flux. These networks have no fixed infrastructure thus making the routing of packets a continuous challenge to be optimally delivered under varying criteria such as when the number of nodes within an area increases the scope for interference between nodes which also increases significantly. Additionally, MANETs have low stability in areas with fast moving nodes which leads to their reduced longevity. This paper proposes a method by adapting the standard Ad hoc On-Demand Distance Vector (AODV) protocol to dynamically adjust transmission power usage, which is titled Dynamic Power-Ad hoc On-Demand Distance Vector (DP-AODV). This method uses the dependence of a transmission range on density to achieve this improvement. The results demonstrate that as density increases, DP-AODV shows decrease in delay than AODV and offering better performance for highly populated networks exceeding 200 nodes. The simulation results show that DP-AODV increase network throughput whilst reducing the node interference in a dense region, as well as it enhances the overall network performance with respect to the increased packet delivery fraction, reducing the control overheads and jitter, enhancing overall throughput, reducing interferences and finally, shortening end-to-end delay in medium to high density conditions. KEYWORDS: MANET, Interference, Density, Delay, Throughput, DP-AODV, AODV. 1. INTRODUCTION The last decade has seen a gigantic rise in the number of wireless network users due to the increasing availability of mobile devices with wireless capability [1,2,3,4]. Besides providing improved mobility for individual users, wireless networks have also allowed organizations to collaborate better, enhance their business flow and increase productivity. Wireless networks [5,6,7,8] are divided into two types: infrastructure and non-infrastructure. For infrastructure networks, base stations plays a key role in coordinating access to one or more transmission channels for mobile devices located within a coverage area and sharing the bandwidth of the wireless channels. GSM and Wi-Fi are the typical examples of such networks. Non-infrastructure networks have no fixed base station or wired getaway and is formed by a collection of mobile nodes on an ad hoc basis. MANETS, in the form of Ad hoc networks and sensor networks are examples of non-infrastructure networks. Ad hoc networks [9, 10] are also known as temporary networks and are used for applications where quick deployment and sharing of information are important. These networks mainly consist of sets of independent mobile nodes that dynamically change their positions so as to automatically establish and maintain routes from sources to destinations. The mobile nodes act both as end systems and also as routers. Crisis management services for disaster recovery and military applications are examples of applications requiring ad hoc networks. The rest of the paper is organized as follows: Section 2 provides an overview of routing in ad hoc, AODV a, Section 3 describes the DP-AODV proposed method, Section 4 discusses the transmission power model, Section 5 describes the simulation environment and the performance metrics, the performance results, and the observation. Finally, the conclusion is provided in Section 6.
1
2. LITERATURE REVIEW In general, the primary purpose of a network routing protocol is to allow packets to be efficiently directed and delivered to their ultimate destinations. Routing protocols for mobile ad-hoc networks have different requirements compared to that of wired networks, and are mainly oriented towards the reduction of the number of transmissions to shorten end-to-end delay optimally rather than any other metric such as throughput. Routing protocols in these types of networks are often classified as either proactive or reactive. Proactive protocols attempt to maintain routes to all nodes in the network, whereas, reactive protocols only generate routes when they are required for transferring data at that instant.
2.1 Ad hoc On-Demand Distance Vector Protocol The AODV [13, 14] algorithm is a reactive routing protocol based on DSDV routing protocol. AODV is hop-byhop in which each intermediate node decides where the routed packet must be forwarded next. As a result, the node in routing contains a route table to maintain new route information that has three important fields: a hop count, next hope node and a sequence number. It introduces low overhead, memory overhead, quick adaptation to dynamic link conditions and low processing. AODV is dependent on the distance vector algorithm. It requests a path when necessary and does not require nodes which are not actively used in connection to maintain routes to the destination. There are some features of AODV, including loop release and link breakages, which lead to the property group of nodes being notified. The algorithm uses individual packets to find and maintain links. Another benefit is the hello message which is broadcast at periodic intervals to the immediate neighbours. AODV has a multicast route invalidation message. AODV routing protocol consists of three different message types: Route Request (RREQ), Route Reply (RREP) and Route Error (RERR). When a node wants to send a data packet (message) to a destination, firstly, it will examine the route table to decide if the route to the endpoint of communication is a valid connection or not. If the route is valid, this node will decide to send a message to the next hop node. If not, it begins route discovery by broadcasting RREQ packet. A RREQ packet consists of a broadcast ID, a source node IP address, the last known sequence number of the destination and a new sequence number for the source. Each node that receives the RREQ should update information for the source node and establish backward pointers to the source node in the route table. The RREQ arrives either at the destination or at an intermediate node from where there is a route to the destination where the RREP is made. Once the real route is made by unicasting a RREP to the origin of the RREQ, every node receiving the RREQ caches a route back to the origin of the RREQ. The intermediate nodes send back RREP to the previous node and become a part of the route to the destination. If the nodes receive the same RREQ packets later, they ignore it and do not send it on. In the case of the source node, it refreshes the entry in the route table and uses this route in the future. Routing of packets is one of the most important challenges facing MANETs due to unstable links, dynamic features and limited resources of these networks [11]. Several attributes of MANETs contribute to the complexity of designing ad hoc routing protocols: density, mobility, device heterogeneity, quality of Service (QoS), battery constraints and bandwidths. One of the main challenges facing mobile communication is transmitting a packet at fixed transmission power that may drain the power consumption of the communication interface. Packets transmitted at a common maximum power trigger a considerable loss on power consumption. Consequently, when node pairs are close to each other, then the power transmission required for them to communicate may be kept at a minimum. In such a scenario, transmitting packets at high power level may generate significant interference to the network and consume more power than required. Thus, an appropriate transmission power for each packet needs to be considered, referred to as a transmission power control. This requires a power level either high enough to guarantee the transmission or low enough to provide less interference and high throughput. Power control minimizes the transmission power by controlling the transmitting range. A dynamic transmission range is more effective in maintaining connectivity while minimizing adverse effects of a high transmission power. Worth noting that power control in ad hoc networks is more important than in cellular networks as the nodes in an ad hoc network 2
communicate by sending data to neighboring nodes. Node density has a great impact on the performance of ad hoc networks in that it influences factors such as routing efficiency, capacity and delay. In 1978, Silvester and Kleinrock published their paper “Optimum Transmission Radii for Packet Radio Networks” [12]. In this paper, the authors explore the trade-off between increased communication radius due to fewer hops used to reach the destination and, the effective channel throughput loss at each node as a result of the increase in transmission range. The paper proves that the best number of neighbours for a given node is 5.89 (rounded to six), and concludes that a node's transmission radius must be adjusted so that it has six neighbours. For a stationary network, the optimum connectivity is seven or eight neighbours per node, which is similar to Kleinrock's conclusion of six neighbours per node [13]. A power control scheme based on the number of neighbours in mobile ad hoc networks was proposed in [14]. In that study, the authors investigated both physical and connectivity design issues. In [15], a mechanism to reduce the transmission and reception power for the frequently used nodes was presented, while another study[16] looked into the transmission power optimization algorithm based on various nearest-neighbor distances algorithm. In [17], the authors proposed an adaptive transmission power control scheme based on the autonomous clustering. In their proposed scheme an estimate of the distance between the current and neighboring node numbers is used to control the transmission power at each node. The work presented in [18] investigates the power management in ad hoc networks by determining the effects of transmission power on network throughput and power consumption. The authors in [18] either allow or disallow nodes to dynamically adjust their power levels per transmission to be the maximum amount of power needed to reach the destination node. In another study [19], the authors present algorithms which adaptively adjust the transmission power of the nodes in response to topological changes. The goal of that study is to maintain a connected network while using minimum power. This paper examines the effect of using different transmission power levels for different number of neighbours by dynamically adjusting the transmission power to minimize the power and to reduce the interference between nodes to achieve a better throughput in the network. AODV protocol has two main characteristics which allow for transmission power improvement: 1)
It relies on dynamically establishing route table entries at intermediate nodes instead of source routing (e.g. DSR) and,
2)
It combines the behavior of DSDV and DSR routing protocols that perform generally well in most cases and are very effective in larger networks.
3. PROPOSED METHOD 3.1 Dynamic Power-AODV Routing Protocol In this paper, we propose a Dynamic Power Ad hoc On- Demand Distance Vector (DP-AODV) protocol which is an improvement of the existing AODV routing protocol. In this extension, we modify all the packets, so that, every packet dispatched from the routing layer will contain the modified packet header with the distance information of the destination and neighbours count. The transmission power is only decided in a wireless physical layer just before transmitting the packet and calculating the transmission power at the wireless physical layer by using the distance information. The hello packets of AODV are modified to carry the “x-y position” coordinate field information. This is done in order to obtain the exact location information of a node which was used by the routing protocol to determine the route. When a node receives the hello message, it calculates the distance for neighbours using the embedded coordinates. Distance of the neighboring nodes is an important aspect in route discovery since all the nodes within the coverage of a particular node can receive the route request messages and process them. Consequently, increasing or decreasing the transmission power of a node increases or decreases respectively the number of one hop neighbours involved in the surrounding domain neighborhood. So, our algorithm is selecting the power needed to maintain the connectivity among the nodes and hence reduce the overall power. Then, each node is added to the neighboring table with the transmission power and distance. Along with the normal information which will be in the AODV routing table, we are maintaining the distances for all the neighboring nodes in the routing table and the routing table entry has not being changed. Every packet is leaving from a node that will contain the distance information about the destination node. So, at the wireless physical layer, while sending the packet to 3
that neighbor node, the transmission power will be changed according to the “level” of power needed to reach the destination node. Fig. 1 shows three scenarios of low, medium to high transmission density in typical dynamic power ad hoc network.
Fig. 1. A Dynamic Power Ad Hoc Network This implies that the selected path will be a stable one based on the densities of the different locations and consequently may not necessarily be the shortest path. The power changing is done in the physical layer and it has a power file to store the transmission power and values in the file. If a packet is scheduled for transmission, it will reach the physical layer (the routing agent will add the distance and the neighbor count inside the packet). While reaching the physical layer, the packet contains the node's neighbor count and the distance. The transmission power can be changed with respect to the neighbor’s count, the distance and the transmitter. DPAODV uses different power levels to determine a route for transmission of a packet. For a small number of neighbours, transmission is done at maximum power; otherwise transmission power is reduced accordingly Fig 1. As our proposal is based on an AODV routing protocol, the fundamental mechanism of data packet handling is the same. However, in our protocol, the basic mechanism of a neighbor based Variable-Power Transmission includes seven steps: 1.
Initially, where each node broadcasts a hello message with its coordinates to determine its neighbours.
Each node uses broadcasting mechanism which sends a packet to every neighbor and the hello packets are generally broadcasted only up to one hop neighbours. But, normally, AODV uses network wide multi-hop broadcast. If a node has already a reduced power to maintain the required number of neighbours, then, the one hop broadcast only reaches up to that distance with respect to that node’s current transmission power. In the present implementation, all the updates in the distance information on the routing table is handled using existing mechanisms of AODV to avoid unnecessary increases in overheads. Each node position is updated dynamically using hello message as shown in Fig 2 with “coordinates” information added as a parameter. Type
Reserve
Destination IP Address Destination Sequence Number Hop Count Lifetime Coordinates Fig. 2. Hello Packet Format in DP-AODV
4
2.
The neighbour node receives a hello message and calculates the distance and adds the entry in the neighbor’s table.
Each node knows its neighbor’s information and the neighbor table contains the neighbour’s_Id, txpower, distance, flag and the neighbor’s count that is used to maintain the neighborhood information. In the Euclidean distance formula, If x = (x1, x2) and y = (y1,y2) then the distance is given by: d (x , y)= √(x2 – x1)2 + (y2 – y1)2 The distance is estimated using Euclidean distance which is only based on the certain coordinates of the nodes that are received through the hello packet. The nodes are sending and updating its location through a hello reply message. When the node knows the distance then that would be the best transmission power level to choose. 3.
Identify the power level for the different number of neighbours.
In our implementation, while transmitting a packet to a destination node, the transmitting node will select the power level based on the number of neighbours and the power required for reaching the destination. 4.
During data transmission the node estimates a number of neighbours by using the neighbour’s count function.
The neighbor count is already available in default AODV through a hello request and reply mechanism. We just add the distance of the neighbor using the same hello mechanism and using the number of entries found in the neighbor’s table.
5.
Transmission power required based on the power level with respect to the number of neighbours.
Collisions within the interference range of a domain/plane are avoided by reducing the transmission range of active neighboring nodes along the route such that they converge to a minimum distance needed to maintain connectivity. The neighboring scheme comprises of four main elements: • • • •
Measurement of the estimated critical range within the domain; Estimation of an ideal node’s power requirement; Allowing a tolerance factor percentage for nodal mobility, some transmission noise and interferences; Selection of the ideal power levels for that node.
When periodic beacons such as the hello packets are received, the node calculates the critical transmission range in the density areas. This is then used to estimate and select the transmission power so that the performance of the routing protocol can be enhanced appropriately. After obtaining the calculated ideal power value, it is periodically checked to ensure that it remains within the designated transmitted power level (Pt) by the receiver. This is to avoid nodes that are currently not involved in the sending of any data packets from adjusting their transmission powers dynamically to a value that is outside the lower and upper bounds to eliminate wastage of energy resources by the participating nodes.
Level 1 If density is low, < 7 neighbours Pt =
Level 2 If density is medium, between 7 and 15 neighbours Level 3 If density is high, > 15 neighbours
5
As an example, in our experiment, we used three different transmission power levels, depending on different number of neighbours involved in the density area range. When the number of nodes is less than 7 neighbours at the source, then Pt = Level 1 in the range which transmits at full power such as 250m. If the number of neighbours is higher than 15 neighbours, then Pt = Level 3 in the range which transmits at a reduced power such as 170m. When a node can be reached only by using the power needed to transmit up to the cross-over distance, then using any power more than that is not necessary. So, at that location, the node will use only a lower level of power to maintain connectivity. Similarly, we use different levels of powers to maintain connectivity at different density areas. 6.
Adjusting the transmission power
Based on the estimated distance transmission power required to transmit the packet to a corresponding neighbor depends on two ray ground propagation. The transmission power is required to reach the next hop or destination. Our algorithm consists of an input parameter, such, as the variable-power. This parameter can adjust the power level based transmission depending on the different number of neighbours. 7.
Forward the packet
It is to transmit the packet using minimum energy required. Each packet handled in the wireless physical layer of the current node will be transmitted as per the number of neighbours of the nodes as well as the next hop is resolved from the routing table (destination). 4. TRANSMISSION POWER MODEL The free space propagation model [23] and the two-ray propagation model [24] are used to calculate the received power of each packet between the neighboring nodes. The free space assumes that the ideal propagation condition of a clear line-of-sight path between the transmitter and receiver exist, and the two-ray propagation assumes that the ground refection path besides the line-of-sight propagation is observed. The relation between the transmitted signal power (Pt) and the received signal power (Pr) is given as:
𝑃𝑟 =
𝑃𝑡 ∗ 𝐺𝑡 ∗ 𝐺𝑟 ∗ 𝜆2 ]1[ (4𝜋)2𝑑2𝐿
Where, Pr : Received Signal Power, Pt : Transmitted Signal Power, Gt and Gr are Transmission and Receiver Gain of Antenna (1.0), d : Distance between the Transmitter and Receiver, dc : Cross Over Distance , L : System Loss, λ : Wavelength. In addition to the line-of-sight path, the two-ray model also takes account of the ground reflection of the path. This model gives a more accurate prediction for longer distances, where hr and ht are Height of antenna for receiver and transmitter (1.5 m). The equation for this model is:
𝑃𝑟 =
𝑃𝑡 ∗ 𝐺𝑡 ∗ 𝐺𝑟 ∗ℎ𝑡2∗ℎ𝑟2 ]2[ 𝑑 4𝐿
If the distance is less than, or equal to, the cross-over distance, then the free-space propagation model is used.
6
Otherwise, the transmission is regarded as a two-ray propagation model. The equation used to calculate the crossover distance is as follows:
𝑑𝑐 =
4𝜋∗ℎ𝑡 ∗ℎ𝑟 𝜆
]3[
Where, dc: Cross Over Distance, which is the reference to the distance of the receiver. Transmission between a transmitting node and a receiving node is only successful if the received power of the radio signal is above a certain threshold. However, if the received power is below this threshold then the transmission is undetected. Moreover, if there are multiple transmissions at the same time, then only the transmission with the highest power is received. The standard NS-2 radio propagation models comprising of the free space propagation model and the two-ray propagation model are used in our simulation. 5. EXPERIMENTAL EVALUATION In our study, we used a Network Simulator 2 (NS-2) [22], which is very popular in the data communication and computer networking technological, research and development community. It is an event-driven simulation tool that is used in order to analyze and evaluate the DP-AODV communication protocol. It supports protocols including AODV, DSDV, DSR, AOMDV and TORA and consists of two main programming languages: C++ and OTcl. 5.1 The Simulation Environment Other than defining the requirements and design of DP-AODV, the methodology and experimentation for implementing the DP-AODV protocol is simulation-based prototyping. It extends the existing AODV protocol by adding some fields in the NS-2 packet header and uses the NS-2 simulator to test and evaluate the enhanced protocol in a variety of scenarios, and to compare the results with AODV routing protocol. Simulation is used to test the protocol as, unlike real-life experiments, it is easy to conduct, less expensive and enables the adjustment of parameters for different scenarios. A Tcl scripting file is an input into the tool, an ASCII trace file corresponding to the event registered at network level and organized according to certain fields and a network animator visualization tool (called NAM) are used for displaying the nodes in the network in real-time, which are also very useful to verify the accuracy of the simulation. NS-2 primarily outputs text-based simulation results. The diagram in Fig. 3 represents the modified modules. C++ is used under NS2 in order to implement the DP-AODV routing protocol, with the use of TCL (Tool Command Language) scripts to describe the simulation scenarios. It shows where our extensions are arranged within the NS2 framework.
7
Fig. 3: Files in the NS-2 Simulator Framework that were modified. The files for the modified modules are as follows: Mac/wireless-phy.cc: NS-2 contains an energy model for wireless nodes that is useful for assessing the advantages of different energy conservation techniques, such as the sleep mode, or using the optimal network densities. The model allows the power requirements for transmitting and receiving the packets, or for an idle mode, to be specified. In Mac/wireless-phy.cc, we have included the routing agent that will add the distance of the destination and the neighbor count inside the packet. Mac/wireless-phy.h: This is the header file where we defined the power timer and power file (double power, double time) to store all the events. Aodv/aodv.cc: It contains, sends and receives hello messages, sends and receives aodv function and the routing table to maintain the information of each node. In this file, all timers, routing agents and Tcl hooks are actually implemented. Aodv/aodv.h: This header file is where we define the mobile node, getDistance (double x, double y), number of neighbours, flag and neighbor tables. We also define the neighbours list (neighbor IDs, txpower and distance). Aodv/packet.h: This is the header file where the AODV reply header (xpos, ypos and transmission power) is declared. Aodv/rtable.h: This is the header file where the routing table and neighbor’s table (txpower and distance) are declared. Common/packet.h: Each packet in NS-2 is associated with a unique type that associates it with the protocol to which it belongs, such as TCP , ARP , AODV , FTP , etc. We defined x, y, txpower, num of nbrs in the packet.h header file. Common/packet.cc: This file contains the size of a packet’s header, free list, off- set of common header and offset of flag’s header, which is accessible through Tcl. It is used to manage active packet header types. Each packet in NS-2 is used to exchange information between objects in the simulation. We defined num of nbrs in the packet.cc source file. The structure of the protocol was created using an agent, which represents the endpoints and can be used to implement the protocol at various layers. The agent is the principal class for implementing the protocol and provides
8
a link with the Tcl interface, for control of the protocol using TCL scripts. New control packet parameters are defined by DP-AODV, in the common packet.h header file, to represent the format of the control packets. The protocol can send packets periodically or following a delay after the occurrence of an event. We also used random scenarios that have different parameters each of which have an effect on the ad hoc movement pattern. The parameters which can be altered are: Maximum speed: a speed is assigned randomly between (0-30m/s) every time. Topology: different number of nodes (75,100,150, 200).
A dynamic network is when the nodes move continuously throughout the simulation period, which is the worst case scenario for the network performance. A static network has very low mobility where nodes are completely static. Initial energy for each node is set at 100 Joules at the start of every simulation in each scenario, and a node will consume 0.9W for sending packets and will consume 0.6W for receiving packets. The maximum speed of the nodes was limited to 30 m/s. The traffic is generated by 40 CBR sources distributing the traffic among all nodes. The sending rate was fixed at 4 packets per second, with a data packet size of 512 bytes. The proposed DP-AODV model assumes random distribution of nodes within a plane or domain in a typical ad hoc network. We used a fixed area in the simulation to ensure consistency and subsequent comparability to the standard AODV protocol. The simulation parameters which are used for the node movement scenario are shown in Table I. Table I. Movement Node Parameters Parameter Value NS-2.34 Simulator version Random waypoint Mobility model IEEE 802.11 Physical/MAC layer DP-AODV and Protocols AODV Free space & TwoPropagation ray Models Omni-Antenna Antenna Model Random Node-placement 2 (Mbps) Bandwidth 1000×1000 (m) Simulation area 100 (sec) Simulation time 75,100,150 and 200 Node Density 0~30 (m/s) Speed 40 (sec) Pause time 10 Iteration Traffic is generated using a constant bit rate from the source nodes. The traffic pattern numbers used by the source nodes are listed in Table II.
9
Table II. Traffic Pattern Parameters Parameter Traffic type Packet size Max connection Max sending packet Initial energy setting Txpower of nodes Rxpower of nodes
Value CBR (UDP) 512 40 4 packets/sec (2 Mb/sec) 100 Joules per node 0.9 (W) 0.6 (W)
5.2 PERFORMANCE METRICS In order to evaluate the performance, several metrics can be used, however for the purposes of this research the following metrics were given for routing protocols evaluation: Packet Delivery Fraction (PDF)
The ratio of received packets by CBR, sink at destination over sent packets by the constant bit rate source (CBR, application layer). The metric tells us how reliable the Protocol is. End-to-End Delay
The cumulative delay that might come about as a result of buffering during discovery of routes queuing at interfaces, delays in retransmission at the MAC, and the time taken for propagation and transfer. Throughput
The amount of data that is transmitted per unit time, (i.e., data bytes delivered to their destination per second. Control Overhead
The ratio of the total number of control packets to the total number of packets sent to the MAC layer. Jitter
Measures the variation in transmission time of packets received at the destination, caused mainly by network congestion, delivery time delays, timing drift or route changes to reach the destination. This mainly happens when a system is in a nondeterministic state. Worth mentioning, the most important metrics for best-effort traffic are the packet delivery fraction and the endto-end delay. However, these metrics are not entirely independent as a shorter delay does not necessarily mean a higher packet delivery fraction, as only successfully delivered packets are used to measure the delay. A lower packet delivery fraction and longer delay may, however, cause a larger overhead. 5.3 PERFORMANCE RESULTS The results of the analysis carried out offer a clear demonstration of the advantages of the improvements made in DP-AODV in comparison with basic AODV. The comparison of the results from the analysis is more realistic and accurate as both the routing protocols used single-path transmission in the experimental scenarios. Varying movement patterns and traffic patterns scenario files were run with both routing protocols in order to ensure a fair comparison without bias. 10
From Fig 4, it can be seen that the packet delivery fraction has always improved in different kinds of densities with the use of DP-AODV compared to standard AODV. It has improved from 12 % to 31% because the proposed method restricts the density by controlling the transmission power. The dropped packet is high in basic AODV than DP-AODV, and many packets are lost due to fixed transmission power and it makes an extensive interference between nodes. As can be seen from Fig 5, the delay has reduced in most of the densities with DP-AODV. It has decreased to 52%. Hence, for routes having a higher success probability and that use less time to transmit data, DPAODV is generally less than that of basic AODV. These results occur as the proposed DP-AODV restricts the density by controlling the transmission power, but with AODV many packets are lost due to fixed transmission power and extensive interference between nodes.
PAKET DELIVERY RATIO (%)
30 m/s Speed 100 80 60 40 20 0 75
100
150
200
DENSITY DP-AODV
AODV
Fig. 4. Packet Delivery Fraction (PDF) for various node densities
30 m/s Speed DELAY (S)
2 1.5 1 0.5 0 75
100
150
200
DENSITY DP-AODV
AODV
Fig. 5. Delay for various node densities Fig 6 shows that DP- AODV gives a shorter EED as compared to AODV because in a dynamic network as the density increases, the EED for DP-AODV increases, though there is a decrease from 150-200 nodes. Likewise, with AODV, as the density increases, the EED increases. Similarly, in a static network, as the density increases, the EED for DP-AODV increases. Likewise, with AODV, as the density increases, the EED increases. The reason for this is that finding a routing with a higher success possibility (as observed for DP-AODV) will definitely use less time to send data from the source node to the destination node.
11
Fig. 6: EED vs Density (Pause Time 0 sec - Dynamic) It can be seen in Fig 7, the throughput of the networks between DP-AODV and basic AODV increases the density. The throughput of the network in DP-AODV is much more enhanced than the basic AODV. The reason is due to the reduction of bandwidth waste by RREQ data packet in the route discovery. From Fig 8, it may be seen that the overhead of the DP- AODV is less than the basic AODV because it sends out less number of control packets that can reduce the overhead in DP- AODV. AODV has a high overhead rate due to use of a large share of network resources. The Fig 5 shows that the delays between packets decrease in DP-AODV and perform well despite node density, while a basic AODV performs poorly in terms of link breakage and jitter conditions as seen in Fig 9. The density performance improves as nodes density within the region increases. When we reduce the transmission level with respect to the distance of the next hop to the final destination then there is no possibility of frequent link failure. For an optimum level of power usage during multi-hop forwarding of packets, each region attempts to use minimum power to decrease the overall communication overhead and invariably reduce power consumption. Generally, DPAODV exhibits better performance and is more efficient than AODV in high density networks as node density gradually increases. In low density, AODV performs better but there is high probability for some errors to occur during the routing discovery process when using the default transmission power. In a dense AODV network, there may be more collisions between neighboring nodes, which means that more energy is wasted at the constant radio level, which is avoided in DP-AODV.
THROUGHPUT (KBIT/S)
30 m/s Speed 250000 200000 150000 100000 50000 0 75
100
150
200
DENSITY DP-AODV
AODV
Fig. 7. Throughput for various node densities
12
CONTROL OVERHEAD (%)
30 m/s Speed 150000 100000 50000 0 75
100
150
200
DENSITY DP-AODV
AODV
Fig. 8. Overhead for various node densities
30 m/s Speed JITTER (SEC)
0.15 0.1 0.05 0 75
100
150
200
DENSITY DP-AODV
AODV
Fig. 9. Jitter for various node densities In summary, the results have shown that DP-AODV performed better than basic AODV across all five of the metrics analyzed. DP-AODV gave a higher packet delivery fraction rate than AODV, particularly in a dynamic environment where the speed of nodes and number of sources where varied. DP- AODV also gave a shorter end-to-end delay than AODV across all metrics, with a typically lower end-to-end delay in a static environment, though there were some minor differences in dynamic environment experiments. The throughput results were also higher for DPAODV, particularly in a static environment when varying the speed of nodes and number of sources. The power consumption results revealed that DP-AODV has lower power consumption than basic AODV, and that power consumption for both protocols was lower in a static environment, with the exception of DP-AODV when the number of sources was varied. The power consumption for DP-AODV remained constant when the pause time was varied. With the final concluding remark that the proposed DP-AODV enhances the overall network performance with respect to increased packet delivery fraction, reduced control overheads and jitter, enhanced overall throughput, reduced interferences and finally, shortened end-to-end delay in medium to high-density conditions, which has significant impact in ad hoc networks. 5.4 FURTHER OBSERVATION Density is a key aspect of ad hoc networks, since high density can cause collision and congestion due to fixed transmission power which leads to high loss of packets in the networks. Another key aspect is the mobility. Since high mobility may cause frequent direct line breaks which could lead to unnecessary re-route discovery, hence waste of a lot of power. The expected outcomes of dynamically changing the transmission power scenario typically means that 13
a node initially increases its transmission power until it detects a set of neighbours in its vicinity and adjusts its power according to the number of neighbours. As the transmission power increases, the number of neighbours increases, and as the transmission power decreases, the number of neighbours decreases, enhancing the network throughput. Ultimately, the transmission power varies between minimum and maximum values. Although there has been considerable research in the field of power varying protocols, the majority of this type of power management protocol, for example COMPOW [25], aims to conserve power and do not specifically address the idea of dynamically reducing interference in the wireless network. However, the enhancements made to the existing AODV protocol by the DPAODV are intended to reduce mutual interference of radios as this directly impacts on energy consumption within the network. The dynamic variation of transmission power usage in order to reduce interference has recently gained in importance within the field of power management research as 3rd Generation Partnership Project (3GPP) standards organization which develops protocols for mobile telephony and networks has defined "5G NR" (5G New Radio) software as "5G". As mentioned before, the results demonstrate that the enhanced protocol (DP-AODV) generates better performance results as compared with a basic AODV. DP-AODV has different characteristics to AODV as it has a higher packet delivery fraction and throughput, a shorter end-to-end delay, lower power consumption and minimum jitter. These key characteristics are the results of both the power control mechanism and the enhanced hello message exchange mechanism that provide both a better success rate for packet delivery and demonstrate that the enhanced algorithm is more energy efficient, which is a major issue for ad hoc networks. The improved performance of DP-AODV compared to AODV can be attributed to several design factors. One of the major factors is the incorporation of the enhanced hello mechanism, which lowers the rate of problems in the algorithm to search for neighboring nodes. Nodes use hello messages to dynamically update the routing information which ensures a more stable link, without unnecessary link breakages, and as a result more successful packet delivery to destination nodes. 6. CONCLUSION The current paper proposed a technique that considered a number of neighbours and a level of power and transmission mechanism, which solved several problems in the network by improving an exciting AODV routing protocol. Our proposed technique evolved around the power level which dynamically increased or decreased, based on node density that was far more effective than fixed transmission power on network throughput. The proposed algorithm automatically adjusted the transmission power at each node so as to keep its number of neighbours within a specified range. Consequently, this lead to the prevention of unwanted interference and unnecessary overhearing and overprocessing by other nodes and thus increased the overall throughput in the network. The simulation results showed that DP-AODV enhanced protocol offered better network performance; hence improved the packet delivery, throughput and reduced interference as compared to AODV routing protocol. In our future work, we will attempt to improve the performance of DP-AODV routing protocol, by finding the appropriate threshold, the appropriate power level to use and investigate the consumed power involving a larger number of wireless nodes and more types of ad hoc network topologies to find a new metrics and measurement that will combine the delivery ratio and the energy consumption. We also expect to decrease the overhead, interference between nodes, and improve the throughput in the network to be adapted in high density and high mobility networks. In addition, we also will attempt to implement DP-AODV future wireless networks 5G cognitive radio networks and IEEE 802.16 based mesh networks in the context of Internet of Things (IoT).
14
Declaration of interests The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
7. REFERENCES [1] Munisha Devi and Nasib Singh Gill. Mobile Ad Hoc Networks and Routing Protocols in IoT Enabled Smart Environment: A Review. Journal of Engineering and Applied Sciences, 14(2019) 802-811. [2] Guaya-Delgado, L., Pallarès-Segarra, E., Mezher, A.M. et al. A novel dynamic reputation-based source routing protocol for mobile ad hoc networks. J Wireless Com Network 2019, 77 (2019) doi:10.1186/s13638-019-13757 [3] Jhaveri, Rutvij H., et al. "Sensitivity analysis of an attack-pattern discovery based trusted routing scheme for mobile ad-hoc networks in industrial IoT." IEEE Access 6 (2018) 20085-20103. [4] S. L. Meshram and P. D. Dorge, "Design and performance analysis of mobile Ad hoc network with reactive routing protocols," 2017 International Conference on Communication and Signal Processing (ICCSP), Chennai, (2017) 0443-0447.doi: 10.1109/ICCSP.2017.8286396 [5] Ejmaa, A. M. E., Subramaniam, S., Zukarnain, Z. A., & Hanapi, Z. M. Neighbor-based dynamic connectivity factor routing protocol for mobile ad hoc network. IEEE Access, 4 (2016) 8053-8064. [6] Han, Nguyen Dinh, Yonghwa Chung, and Minho Jo. "Green data centers for cloud-assisted mobile ad hoc networks in 5G." IEEE Network 29 (2015) 70-76. [7] H. Xiao, D. M. Ibrahim and B. Christianson, "Energy consumption in Mobile Ad Hoc Networks," 2014 IEEE Wireless Communications and Networking Conference (WCNC), Istanbul, (2014) 2599-2604. doi: 10.1109/WCNC.2014.6952818 [8] CR Komala, P.S. Srinivas Shetty, and E. Elevarasi, “Wireless Ad Hoc Mobile Networks,” In National Conference on Computing, Communication and Technology, pp. 168–174, 2010. [9] Mobile Ad hoc Networks (MANET). (2012). [Online]. Available: http://datatracker.ietf.org/wg/manet/charter/, IETF Working Group. [10] A.H.A. Rahman and Z.A. Zukarnain. “Performance Comparison of AODV, DSDV and I-DSDV Routing Protocols in Mobile Ad Hoc Networks,” In European Journal of Scientific Research, vol. 31, no. 4, pp. 566– 576, 2009. [11] A. Boukerche, B. Turgut, N. Aydin, M.Z. Ahmad, L. Boloni, and D, “Turgut. Survey paper: Routing Protocols in Ad Hoc Networks: A Survey Computer Networks,”: In the International Journal of Computer and Telecommunications Networking, vol. 55, no. 13, pp. 3032–3080, 2011. [12] L. Kleinrock and J. Silvester, “Optimum Transmission Radii for Packet Radio Networks or Why Six is a Magic Number,” In Proceedings of the IEEE National Telecommunications Conference, vol. 4, pp. 1– 4, Birimingham, Alabama, 1978. [13] E.M. Royer, P.M. Melliar-Smith, and L.E. Moser, “An Analysis of the Optimum Node Density for Ad Hoc Mobile Networks,” In International Conference on Communications, vol. 3, pp. 857–861, IEEE, 2001. [14] B. Abasgholi, R. Kazemi, M. Arezoomand, and A.R. Enayati, “Neighbor Detection Power Control (NDPC) MAC Protocol in Mobile Ad Hoc Networks,” In 3rd International Symposium on Wireless Pervasive Computing., pp. 508–512, IEEE, 2008. [15] H. Abusaimeh and S.H. Yang, “Reducing the Transmission and Reception Powers in the AODV,” In International Conference on Networking, Sensing and Control, pp. 60–65, IEEE, 2009. [16] Y. Chen, H. Yang, B. Liu, and J. Cheng, “Transmission power optimization algorithm in wireless ad hoc networks,” In International Conference on Communications and Mobile Computing., vol. 3, pp. 358- 363, IEEE, 2010. [17] R. Oda, T. Ohta, and Y. Kakuda, “A transmission power control adaptive to variation of node density in mobile ad hoc network clustering,” In Second International Conference on Communication Software and Networks, 15
pp. 141–146, IEEE, 2010. [18] T.A. ElBatt, S.V. Krishnamurthy, D. Connors, and S. Dao, “Power Management for Throughput Enhancement in Wireless Ad-hoc Networks,” In IEEE International Conference on Communications., vol. 3, pp. 1506–1513, IEEE, 2000. [19] R. Ramanathan and R. Rosales-Hain, “Topology Control of Multihop Wireless Networks using Transmit Power Adjustment,” In Nineteenth Annual Joint Conference of the IEEE Computer and Communications Societies, Proceedings., vol. 2, pp. 404–413, IEEE, 2000. [20] I.D. Chakeres and E.M. Belding-Royer, “AODV Routing Protocol Implementation Design,” In 24th International Conference on Distributed Computing Systems Workshops. Proc., pp. 698–703, IEEE, 2004. [21] C.E. Perkins, E. Royer, and S. Das, “Ad Hoc on Demand Distance Vector (AODV),” Request for Comments (RFC), 2003. [22] The Network Simulator ns-2. [Online]; Available: http : // www.isi.edu/nsnam/ns/. [23] C. S. R. Murthy and B. Manoj, Ad hoc wireless networks: Architectures and protocols. Prentice Hall, 2004. [24] C. E. Perkins, Ad hoc networking, vol. 1. Addison-wesley Reading, 2001. [25] S. Narayanaswamy, V Kawadia, R. S. Sreenivas and P. R. Kumar, “The COMPOW protocol for power control in ad hoc networks: Theory, architecture, algorithm, implementation, and experimentation,” In European Wireless Conference, pp. 156-162, 2002.
16