An optimized QoS-based clustering with multipath routing protocol for Wireless Sensor Networks

Journal of King Saud University – Computer and Information Sciences xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

Journal of King Saud University – Computer and Information Sciences journal homepage: www.sciencedirect.com

An optimized QoS-based clustering with multipath routing protocol for Wireless Sensor Networks O. Deepa a,⇑, J. Suguna b a b

Research & Development Centre, Bharathiar University, Coimbatore, TN, India Department of Computer Science, Vellalar College for Women, Erode, TN, India

a r t i c l e

i n f o

Article history: Received 1 August 2017 Revised 19 October 2017 Accepted 29 November 2017 Available online xxxx Keywords: Energy hole problem OQoS-CMRP Modified PSO-based clustering algorithm SingleSink-AllDestination algorithm Round-robinPathsSelection algorithm Wireless Sensor Networks

a b s t r a c t As the Wireless Sensor Networks (WSNs) continue to evolve, it becomes more and more significant in day today life. WSNs are a promising approach used in various applications but finding an optimal route discovery is more problematic due to dynamicity, heterogeneity, resource scarcity and so on. Generally, residual energy of sensors in sink coverage area is drained very quickly compared to other areas in WSNs. The proposed Optimized QoS-based Clustering with Multipath Routing Protocol (OQoS-CMRP) for WSNs reduces the energy consumption in sink coverage area by applying the Modified Particle Swarm Optimization(PSO)-based clustering algorithm to form clusters to select cluster heads in the sink coverage area and to solve energy hole problem. The SingleSink-AllDestination algorithm is used to find near optimal multi-hop communication path from sink to sensors for selecting the next hop neighbor nodes. The Round-robinPathsSelection algorithm is used for transferring data to sink. According to QoS metrics, the performance of the proposed communication protocol is evaluated and compared with other existing protocols namely EE-LEACH and EPSO-CEO. The simulation result shows that OQoS-CMRP for WSN achieves prominent data communication with reasonable energy conservation. It also reduces transmission delay and communication overhead on the basis of ensuring the outcome of the entire network. Ó 2017 The Authors. Production and hosting by Elsevier B.V. on behalf of King Saud University. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Contents 1. 2.

3.

4.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Related work. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Routing protocols for WSNs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Overview of PSO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.1. PSO based routing protocol for WSN. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . OQoS-CMRP problem formulation for WSNS model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Network environment and assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Metric description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.1. Reliability metric . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.2. Energy metric . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.3. Delay metric . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Proposed OQoS-CMRP for WSNs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

00 00 00 00 00 00 00 00 00 00 00 00

⇑ Corresponding author. E-mail addresses: [email protected] (O. Deepa), [email protected] (J. Suguna). Peer review under responsibility of King Saud University.

Production and hosting by Elsevier https://doi.org/10.1016/j.jksuci.2017.11.007 1319-1578/Ó 2017 The Authors. Production and hosting by Elsevier B.V. on behalf of King Saud University. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Please cite this article in press as: Deepa, O., Suguna, J. An optimized QoS-based clustering with multipath routing protocol for Wireless Sensor Networks. Journal of King Saud University – Computer and Information Sciences (2017), https://doi.org/10.1016/j.jksuci.2017.11.007

2

O. Deepa, J. Suguna / Journal of King Saud University – Computer and Information Sciences xxx (2017) xxx–xxx

5.

6.

4.1. Cluster formation phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. Route discovery phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3. Data transmission phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4. Rerouting or Re-clustering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1. Simulation results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2. Performance analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.1. Network lifetime. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.2. Total energy consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.3. Average residual energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.4. Throughput . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.5. Packet Delivery Ratio (PDR) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.6. Normalized Overhead. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.7. End-to-End delay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion and future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1. Introduction In recent years, Wireless Sensor Networks (WSNs) have become one of the developing research field, as they are envisioned to have wide applications with different phenomenon related to environmental tracking, emergency response, security monitoring in manned or unmanned missions (Akyildiz et al., 2002; Romer and Mattern, 2004; Akkaya and Younis, 2005). A WSN is composed of huge, low power intelligent sensors with high power sink, which are responsible for establishing paths among themselves with certain transmission regulations (Yick et al., 2008). Wireless sensors are more advantageous due to their simple installation, selfidentification, self-diagnosis and time awareness for coordination with other sensors to form dynamic self-organized networks. But, it is tightly constrained with limited energy, memory, analytical and computational ability and also has low data rate as well as short range for wireless radio transmission (Toldan and Kumar, 2013). Routing is a main issue in WSNs. Traditional routing protocols cannot be applied to sensor networks directly because sensor network differ from other ad hoc networks in terms of the following perceptions namely battery-operated sensors, lightweight routing protocols and adaptive communication patterns (Villalba et al., 2007). Since WSNs have an ad hoc topology and there is no infrastructure, finding a path and transmitting data to the sink is an interesting and critical task (Sohrab et al., 2000). For example, the reliable and delay sensitive applications like patient monitoring, fire detection, gas leakage monitoring and homeland security, the sensory data should be transmitted within a specific delay and reliable data communication are highly essential. Therefore, enabling such applications in WSNs require QoS and energy awareness which is to be considered in different layers of the protocol stack (Heinelman et al., 2002). The basic idea is to reduce the load of the sensor by giving more responsibility to the sink and to propose the lightweight clustering with multipath routing protocol for WSN by considering the multiconstrained QoS metrics. The protocol is named as Optimized QoSbased Clustering with Multipath Routing Protocol (OQoS-CMRP). In WSN, sensors nearby the sink always transmit huge data; as a result they die. Finally, the network is partitioned and the sink cannot receive any data. This occurrence is known as energy hole problem or hot spot problem (Sharma and Lobiyal, 2015; Mohemed et al., 2017). This protocol conserves sensor’s energy in the sink coverage area by applying Modified PSO-based clustering technique to reduce the energy hole problem, conserve the residual energy of sensors and also enhance lifetime of sensor network especially in the sink coverage area. The process of selecting the cluster head selection and finding optimal routes are assigned to

00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00

the sink because of the unlimited resources at the sink. Clustering has been proven to be one of the efficient techniques for saving energy of the sensor networks. For designing any clustering scheme, the network is divided into numerous groups, called clusters. Each cluster has a leader denoted as cluster head. Within the clusters, cluster heads are responsible to collect the data from their member sensors. Cluster head performs data aggregation and removes the redundant data, thus reduces energy depletion of the network. However, cluster heads consume more energy due to additional overload for receiving and aggregating the data and transmitting the data to the sink. It also has the effect on the routing process in an energy efficient method which is the vital goal of any WSN. Hence, the proper selection of cluster heads plays an important role for conserving the energy and enhancing the lifetime of WSNs. Cluster head selection is an optimization problem which is NP-hard. Particle Swarm Optimization (PSO) is swarm intelligence based optimization method, which can be a better choice for selecting cluster head due to its ease of implementation, global search, ability to escape from the local optima and rapid convergence for the globally best solution. Furthermore, the design of routing protocol is resilient to frequent path disruptions caused by node or link failure and collision. Some of the routing protocols find routing path but often fails during data transmission, which decreases the reliability. The solution to this problem is designing an enhanced multipath routing protocol. Multipath routing protocol allows numerous paths between the source and the sink. So if one of the paths fails, data can be sent through an alternative path. This process enhances the delivery ratio with minimum latency of the network. This motivated to employ the concept of multipath routing for reliable transmission, which reduces the control overhead for route discovery and increases the throughput of the system. In order to establish the near optimal communication path between the sink and sensors, it must start a route discovery algorithm which is required for selecting the next hop node in the sensor network based on a topological structure and QoS metric values of neighbor nodes. If an event occurs, source sensor transfers the data to the sink across various best-case multipath in the fixed time slice period by using Round-robin Paths Selection algorithm, which distributes the traffic load. Thus, the proposed routing protocol may provide better data throughput while minimizing packet loss with minimum delay and enhance the lifetime of the network. The rest of this paper proceeds as follows. Section 2 briefly reviews background information and the works related to the design and overview of PSO. Section 3 presents network model and QoS metrics. Section 4 proposes an OQoS-CMRP to optimize the performance of clustering and network routing mechanism using Modified

Please cite this article in press as: Deepa, O., Suguna, J. An optimized QoS-based clustering with multipath routing protocol for Wireless Sensor Networks. Journal of King Saud University – Computer and Information Sciences (2017), https://doi.org/10.1016/j.jksuci.2017.11.007

O. Deepa, J. Suguna / Journal of King Saud University – Computer and Information Sciences xxx (2017) xxx–xxx

PSO-based clustering algorithm and SingleSink-AllDestination algorithm respectively. In section 5, the simulation environment is constructed and the experimental results are discussed. Finally, this paper summarizes the conclusion and future work.

2. Related work 2.1. Routing protocols for WSNs The routing for WSNs is tough problem due to the inherent characteristics that distinguish from other wireless networks. Routing protocols designed for WSNs which differ from different perception depending on their application, architecture and goal (Amit and Senthil Murugan, 2016). Ganaesan et al. (2001) have presented a disjoint multipath routing, which is a distributed algorithm based on local information and achieve load balancing. This algorithm uses a primary path to transfer data. An alternative path can be used when the primary path fails. However, this algorithm is not suitable for extending lifespan of the network. De et al., (2003) have proposed a meshed multipath routing with an efficient strategy. This algorithm achieved a better throughput than the traditional multipath algorithms. However, this approach needs nodes to be equipped with Global Positioning System (GPS), which maximizes the cost of the node. Younis and Fahmy (2004) have presented a clustering protocol named as Hybrid, Energy-Efficient Distributed (HEED), which prolonged the lifespan of the sensor network. This protocol has formed the clustering and cluster head selection based on the node degree, communication cost and residual energy of sensors. It operates in multi-hop networks, using an adaptive transmission power in the inter-clustering communication. However, the cluster selection deals with only a subset of parameters, which can possibly impose constraints on the system. Yang et al., (2010) have proposed a Multipath Routing protocol Based on the credible Cluster Heads (MRBCH) that taken into consideration for both energy efficiency and security. This protocol selects the high energy node as cluster head and then authenticates it by the neighbor cluster heads. Using trust mechanisms, the credit value is generated and exchanged among the cluster heads exclusively. By using multipath cluster head routing based on the credit value ensures a high quality route and prolong the lifespan. Arumugam and Ponnuchamy (2015) have introduced an Energy-Efficient Low Energy Adaptive Clustering Hierarchy (EE-LEACH) protocol, which improved the lifespan and data ensemble of the network. The coverage probability is derived with respect to the Gaussian distribution. In this system, a cluster head is elected for each cluster, which has the maximum residual energy. Also, conditional probability theorem is used for node aggregation. It provide better packet delivery ratio with lesser energy utilization. But it lacks to provide the confidentiality and integrity of data. In QoS based routing protocols, the sensor network has to balance between energy consumption and data quality (Bhuyan et al., 2010). Sohrabi et al., have presented a Sequential Assignment Routing (SAR) (Sohrabi et al., 2000), which uses tree architecture to route data packets in multiple paths. This algorithm takes routing decision based on the QoS metrics on each path, the priority level of each packet and energy resource. To transmit data to sink, SAR computes a weighted QoS metric, which is the product of the additive QoS metric and a weight coefficient associated with the priority level of that packet. Although, SAR ensures energy efficient, fault-tolerance and easy recovery but suffer from large scalability of sensors. The protocol must periodically recalculate the routes to be prepared in case of topology changes due to node failure. He et al. (2003) have proposed a protocol SPEED. This is another

3

location based QoS routing protocol, which provides soft realtime end-to-end guarantee with a desired delivery speed across sensor networks. Felemban et al. (2006) have presented a Multipath and Multi-speed routing protocol (MMSPEED), which maintenances various speed services and probabilistic QoS guarantee in order to avoid congestion and decrease the packet loss rate. This protocol is scalable for large networks, but the energy metric is not taken into consideration. Huang and Fang (2008) have proposed a Multi-Constrained QoS Multipath routing Protocol (MCMP), which improves the network performance with reasonable energy consumption. This protocol has used braided multipath to send data packets according to QoS metrics namely reliability and delay. The problem of end-to-end delay is solved by using linear integer programming. However, this protocol routes the data over the minimum hop count path to satisfy the required QoS, which leads in some cases to consume more energy. Alwan and Agarwal (2013) have presented a Multi-objective QoS Routing (MQoSR) protocol. In MQoSR, the requirements of an application are modeled as multiple QoS classes in terms of reliability, delay and energy. In particular, the problem of providing QoS routing is formulated by using a heuristic neighbor selection mechanism that uses the geographic routing mechanism combined with the QoS requirements to provide MQoSR for different application requirements. Bagheri and Ghaffari (2011) have introduced a Reliable and Energy effective Clustering based Multipath routing algorithm for WSN, in which sensors are equipped with GPS system. The cluster head selection is based on the residual energy of the sensor. The multipath is constructed through the cluster heads and selects another path, if it fails. The control packet overhead of this routing protocol is high, which leads to more energy consumption and it directly affect the network lifetime. Mazaheri et al. (2012) have presented a QoS base multipath hierarchical routing. Among the sensors in the range r select the cluster head based on the residual energy and the distance from the sink. For multipath construction, cluster head chooses the set of cluster heads within the range RðR > rÞ based on the energy, distance, signal to noise ratio and remaining buffer size. This protocol gives more importance on reliability but neglect some QoS parameters such as control overhead, end-to-end delay and network lifetime. Almalkawi et al. (2012) have presented a cross layer based clustered multipath routing. The sink initiated the cluster formation by broadcasting the control packet and based on received signal strength, the powerful nodes become the cluster heads. The cluster heads are classified in different levels. They send the data through the upper level cluster head. This protocol decreases the reliability of the networks because it did not maintain the proper path. Sharma and Jena (2015) have introduced an energy efficient routing scheme using the clustering and multipath technique. The workloads of the sensors are alleviated by giving more responsibility to the sink. The multipath routing technique gives more reliability to the network and it increases the throughput and decreases latency. In addition to that, cluster based data collection reduces the traffic and energy consumption and also increases the network lifetime. 2.2. Overview of PSO Swarm intelligence is becoming one of the modern research fields of computational intelligence especially with regard to selforganizing and decentralized systems. Swarm intelligence simulates the behavior of animal or human populations. PSO is a population-based swarm intelligence technique to solve discrete and continuous problems which improve the computation effectiveness, produce high quality solution and many variants have been proposed (Kennedy and Eberhart, 1995; Del Valle et al., 2008). PSO is inspired from the nature social behavior and dynamic

Please cite this article in press as: Deepa, O., Suguna, J. An optimized QoS-based clustering with multipath routing protocol for Wireless Sensor Networks. Journal of King Saud University – Computer and Information Sciences (2017), https://doi.org/10.1016/j.jksuci.2017.11.007

4

O. Deepa, J. Suguna / Journal of King Saud University – Computer and Information Sciences xxx (2017) xxx–xxx

movements with communications of insects, bird flocking or fish schooling i.e., how they can exploit and explore the search space for shelter and food. In problem space, PSO encompasses with a population ðN p Þ called a swarm of each candidate solution and is represented as a particle ðP i ; 1 6 i 6 N p Þ. Iteratively, all of the particles move around the problem space to find the position, X i;d , according to velocity, V i;d ; 1 6 d 6 D in the dth dimension of the search space. In the initialization of PSO, each particle is assigned with a random position and velocity to move in the search space. The particle moves to a best position based on its individual knowledge and the knowledge gained by the swarm. In the D-dimension problem space, to indicate its current position and velocity of the ith particle at kth iteration is denoted as X i ðkÞ ¼ fxi1 ðkÞ; xi2 ðkÞ; . . . ; xiD ðkÞg and V i ðkÞ ¼ fv i1 ðkÞ; v i2 ðkÞ; . . . ; v iD ðkÞg respectively. For every iteration, each particle fitness value is evaluated and chooses the best position of each particle based on fitness value. In kth iteration, the best position of the particle i is represented by P i ðkÞ ¼ fpi1 ðkÞ; pi2 ðkÞ; . . . ; piD ðkÞg as the personal best. The position of all the particles which give the best fitness value at iteration k is also stored as the global best position denoted byGðkÞ ¼ fg 1 ðkÞ; g 2 ðkÞ; . . . ; g D ðkÞg: 2.2.1. PSO based routing protocol for WSN Latiff et al. (2007) have introduced an energy-aware cluster head selection using PSO called PSO-C by considering various metrics such as average intra-cluster distance and ratio of total initial energy of all sensors to the total current energy of all cluster heads. The cost function of this protocol is to minimize both the maximum average Euclidean distance of sensors to their related cluster heads. But, it does not consider the distance of sink, for the direct communication of cluster heads to sink. The sink distance plays a vital role to reduce the energy consumption of the network. Moreover, it assigns the non-cluster head nodes to the nearest cluster head in the cluster formation phase, which may cause the energy inefficiency of the network and may reduce the lifetime of the network. Singh et al. (2012) have introduced a novel energy-aware CH selection algorithm using PSO, which present the fitness function based on node density, residual energy and distance. But, it does not consider the cluster formation, which may cause severe energy inefficiency to the network. Riham S. Elhabyan and Yagoub (2014) have proposed a PSO-based hierarchical clustering protocol (PSO– HC) for WSNs. The objective of the protocol is to find the optimal number of cluster heads to minimize the energy consumption. Moreover, the protocol tries to maximize the network scalability and coverage by building two–hop communication between the sensors and their corresponding cluster heads. Srinivasa Rao et al. (2016) have introduced a PSO based algorithm for time sensitive applications. It has taken care of energy efficiency and enhances the network lifetime. But, it does not consider the fault tolerance of the network. Vimalarani et al. (2016) have proposed an Enhanced PSO-based Clustering Energy Optimization (EPSOCEO) algorithm for WSN. Energy conservation is done in each sensor by using PSO-based cluster formation and cluster head selection based on the residual energy and the distance from the cluster member to sink. The above work is inspired to propose an OQoS-CMRP, which is the tradeoff between a certain guaranteed QoS requirements and acceptable computational complexity. 3. OQoS-CMRP problem formulation for WSNS model 3.1. Network environment and assumptions A WSN is modeled as an undirected weighted graph GðV; E; WÞ, where V ¼ fv 1 ; v 2 ; . . . v n g, denotes the finite set of vertices that

represent the sensors, E ¼ fðv 1 ; v 2 Þ; . . . ðv n1 ; v n Þg, indicates the finite set of edges that represent the bidirectional wireless links and W is the weight set of all links. In this paper, we assume that the sensors are heterogeneous, dynamicity and have unique ID. Each sensor or cluster head have uniform transmission radius r and its neighbor nodes or cluster members are randomly distributed within the area of pr2 . Each sensor computes its residual energy level as well as energy depletion which varies depending on the location of a sensor to transfer a bit of data. The radius value of sink is 2r and its maximum communication area is 2Pr 2 . 3.2. Metric description The proposed protocol OQoS-CMRP for WSN mainly focuses on three important QoS metrics in sensor network design – the reliability, the energy consumption and the delay, to build a new clustering with multipath routing strategy over large network (Deepa et al., 2016). The proposed OQoS-CMRP for WSN is formulated as link-based and path-based metrics. Each sensor is able to record the link performance between itself and its neighbor in terms of reliabilityðRlink Þ, energy ðElink Þ, delay ðDlink Þ, distance to sink and hop count. The path-based metrics are represented as ðRpath Þ; ðEpath Þ and ðDpath Þ. In multipath routing, to find the total end-to-end guarantee on all the used paths is divided into endto-end reliability ðRe2e Þ, end-to-end energy consumption ðEe2e Þ and end-to-end delay ðDe2e Þ. 3.2.1. Reliability metric The reliable data transmission is an important key of QoS and is calculated to measure the probability of successful packet delivery and can be stated in terms of the Packet Delivery Ratio (PDR) which can be expressed as

PDR ¼

Psink Psource

ð1Þ

where Psource is the total number of packets generated by the source and Psink is the total number of packets received at the sink. The reliability of a single path Rpathi is considered for transferring data packet from source to sink by

Rpathi ¼

hop Yi

Rlinkj

ð2Þ

j¼1

where hopi is the hop count of pathi ; hop P j P 1. Rreq is an application-specific threshold value which reflects the required path reliability for data transmission, only if the reliability of path that satisfies Rpath P Rreq . The reliability objective function f R is to maximize the data transmission reliability Rpath , such that Rpath P Rreq . The multipath end-to-end reliability Re2e for data delivery is calculated between source and sink and is based on the selected path or paths which is given by

Re2e ¼ 1 

np Y

ð1  Rpathi Þ

ð3Þ

i¼1

where np is the number of paths, np P i P 1 used for data transmission to route the data packet. The network has to satisfy the gained reliability compared to the required reliability by an application. The source codes each data packet of size sF bits it receives into F fragments, each of size s and generates another A coding fragment to have in total a set of F þ A fragments. This set of fragments are then transmitted as packets p1 to pn to the sink, allowing at most A lost fragments and the coding rate is thus defined as A/(F + A). The successful probability of packet received by the sink, P succ , to achieve the

Please cite this article in press as: Deepa, O., Suguna, J. An optimized QoS-based clustering with multipath routing protocol for Wireless Sensor Networks. Journal of King Saud University – Computer and Information Sciences (2017), https://doi.org/10.1016/j.jksuci.2017.11.007

O. Deepa, J. Suguna / Journal of King Saud University – Computer and Information Sciences xxx (2017) xxx–xxx

requested reliability, Rreq , has a binomial distribution that depends on Rpath and can be written as

 FþA  X FþA ½ðRe2e Þ x ½ð1  Re2e ÞFþAx  Psucc ½x P F ¼ x x¼F

ð4Þ

2

power loss) for packet transmission using direct or 4

single-hop communication and the multipath fading (d power loss) channel for the purpose of multi-hop communication. Thus the energy consumption for transferring n bits of data packet over distance d meters is calculated as

(

ET ðn; dÞ ¼

nEelec þ nefs d ; 2

d < d0

nEelec þ nemp d ; d P d0 4

ð5Þ

sffiffiffiffiffiffiffi

efs emp

ð6Þ

The energy consumption for receiving n bits of data is calculated as

ER ðnÞ ¼ nEelec

ð7Þ

Thus the energy level of transferring and receiving n bits of data packet in a link is expressed as

Elink ¼ ET ðn; dÞ þ ER ðnÞ

ð8Þ

The remaining energy of a node is represented as

Ecurrent ¼ Ecurrent  ET ðn; dÞ  ER ðnÞ

ð9Þ

The energy consumption of a path is the sum of the energy expended at each link along the path for transmission of data from source to sink and is calculated by

Epathi ¼

hop Pi j¼1

Elinkj

ð10Þ

To transfer n bits of data packet, energy consumption of a path Epath must be less than or equal to the required minimum energy Ereq . The energy objective function f E that reductions the total energy consumption on a path is expressed by

f E ¼ minfEpath g

n bits b bits=sec

ð14Þ

Propagation delay dpropa measures the time required for a bit to travel from source to sink

dpropa ¼

d meters s meters=sec

ð15Þ

Processing time is the time to select the next hop sensor to transmit the data packet. Queuing time is the delay time taken for each intermediate sensor to hold the data packet before it can be processed. The delay of a path Dlinkj is the sum of all delays at all the intermediate sensors along the path and is calculated as follows

Dpathi ¼

hopi X Dlinkj

ð16Þ

j¼1

where the amplifier energy, efs and emp depends on the transmission distance, Eelec is the electronics energy in transceiver, d is distance between source and destination and the threshold d0 is crossover distance

d0 ¼

ð13Þ

where transmission delay dtrans measures the time between the first bit leaving source and the last bit arriving at sink

dtrans ¼

3.2.2. Energy metric In WSNs, energy optimization is one of the most important system design goals (Kulothungan et al., 2011). An energy model is considered in physical and MAC layer of WSN based on the concept of energy depletion which is directly proportional to the communication distance. Two channel propagation models used are the free space (d

Dlinkj ¼ dtrans þ dpropa þ dproces þ dqueue

5

Therefore, the delay objective function, f D is to ensure that the path delay on the selected single path is the minimum, such thatDpath 6 Dreq , where Dreq is an application-specific threshold value which reflects the required single path delay for data delivery. In multipath routing, the overall end-to-end delay De2e is evaluated based on a number of used paths and is given by

De2e ¼

np X

Dpathi

ð17Þ

i¼1

4. Proposed OQoS-CMRP for WSNs In WSNs, sensors typically operate on non-rechargeable batteries, so effective utilization of energy expenditure prolongs the lifespan of the sensor. In order to fulfill the QoS requirement, the sensor network should be able to maintain certain reliability and delay of the specific application. Therefore, developing new routing schemes for WSN to minimize the end-to-end delay and energy consumption is considered to be major performance criteria. Providing maximum reliability and network lifetime while satisfying the QoS requirements represent a critical problem to be addressed. The system model of proposed protocol OQoS-CMRP for WSN is illustrated in Fig. 1. The proposed protocol is divided into four phases: the first phase establishes cluster formation using Modified PSO-based Clustering algorithm, the second phase finds the

ð11Þ

In multipath routing, the total end-to-end energy consumption Ee2e of all the used paths are given by

Ee2e ¼

np X Epathi

ð12Þ

i¼1

3.2.3. Delay metric In WSNs, most of the applications are delay sensitive, so delay is an important QoS metric for reporting data to the sink with very small latency. Delay is the time elapsed from the departure of a data packet from source to sink. The delay metric between two sensors represented as Dlinkj is given by Fig. 1. System Model of Proposed Protocol OQoS-CMRP for WSN.

Please cite this article in press as: Deepa, O., Suguna, J. An optimized QoS-based clustering with multipath routing protocol for Wireless Sensor Networks. Journal of King Saud University – Computer and Information Sciences (2017), https://doi.org/10.1016/j.jksuci.2017.11.007

6

O. Deepa, J. Suguna / Journal of King Saud University – Computer and Information Sciences xxx (2017) xxx–xxx

shortest optimal path using SingleSink-AllDestination algorithm and the third phase provides the procedure for actual data routing and the last phase is Rerouting or Re-clustering. 4.1. Cluster formation phase In sink vicinity, sensors adjacent to sink mostly act as the relay nodes which deliver the data to the sink. These sensors consume more energy as compared to other nodes that are far from the sink, consequently, they die. It creates hotspots in the sink locality and the network gets isolated. Such situation is called as hotspot or energy hole problem. Modified PSO-based clustering algorithm enhances the lifetime of the network by lessening the hotspot problem. The sink initiates the cluster formation process only in the sink coverage area by using Modified PSO-based Clustering Algorithm. The sink transmits Info_req_msg to all sensors in this coverage area. When receiving the request message, sensors start to send its information via Info_reply_msg which contains Sensor_ID, position, velocity and current residual energy. This information is maintained and updated at the sink. For better understanding of the Modified PSO-based clustering algorithm, the following are terminologies listed below S: The set of particles in the sink vicinity, N ¼ f1; 2; . . . ng C n : The number of cluster members covered within the cluster from the current particle ECP-threshold : The threshold energy for being a current particle dmax : The maximum communication range of a current particle and member sinkmax : The maximum communication range of a sink Eðcurrent particleÞ: The residual energy of current particle Eðmember iÞ: The residual energy of ith member dðcurrent particle; member iÞ: The distance between the current particle and ith cluster member dðcurrent particle; sinkÞ: The distance between the current particle and sink Let us consider a sample problem space in which the number of particles as N in sink coverage area. The fitness function decides which particle has the best value in the swarm and also determines the best position of each particle over time. The proposed fitness function for Modified PSO-based clustering is calculated for each particle by using the equation

f p ¼ a1 x1 þ a2 x2 þ a3 x3

ð18Þ

where a1 and a2 are weighing parameters value between 0 and 1 and a3 ¼ 1  a1  a2 n X

x1 ¼

dðcurrent particle; member iÞ

i¼1

Cn

ð19Þ

n X Eðmember iÞ

x2 ¼

x3 ¼

i¼1

Eðcurrent particleÞ 1 Cn

ð20Þ

ð21Þ

The Eq. (20) states that the energy of the current particle must be greater than the threshold value, which is the average energy of all the members. The constraints are subject to

dðcurrent particle; member iÞ 6 dmax

dðcurrent particle; sinkÞ 6 sinkmax

ð23Þ

Eðcurrent particleÞ P ECP-threshold

ð24Þ

0 < a1 ; a2 < 1

ð25Þ

the Eq. (22) states that the members are within the cluster communication range of current particle. The Eq. (23) states that the current particles are within the maximum communication range of the sink. And, the Eq. (24) states that the energy of the current particle must be greater than the threshold value. Each particle adjusts its travelling speed dynamically corresponding to the flying experiences of itself and its swarm experience. Each particle modify its velocity by

v elocitynew ¼ x  v elocityold þ c1 v1 ðpBest  lBestÞ þ c2 v2 ðgBest  lBestÞ

ð26Þ

where x; 0 < x < 1 is an inertia weight of node velocity, c1 ; c2 ; 0 6 c1 ; c2 6 2 are the acceleration coefficients and v1 ; v2 ; 0 < v1 ; v2 < 1 are the randomly generated values, which are basic PSO tuning weights of node position (i.e., the particle to continue moving in the same direction and with the same velocity). The lBest, pBest and gBest are the current, personal and global best position of the particle respectively. The current velocity of a chosen particle is considered to the rate at which the particle’s position is changed. The new position of the particle is evaluated based on the previous position and the updated velocity value.

positionnew ¼ positionold þ v elocitynew

ð27Þ

Again, calculate the fitness value of the new particle by using fitness function in Eq. (18). Then by comparing the fitness value of old particle with new particle, the best one is selected for the next iteration. For every iteration, one best solution is selected as a pBest. Then, the particle which has maximum fitness value in all iteration is selected as a gBest solution. Suppose, if gBest value is obtained in the ith iteration, fitness values of all particles in that particular iteration are taken into consideration for cluster formation by making the nodes in its radio range as its cluster members. The gBest value is broadcasted to each cluster head so that, each cluster head may aware of the gBest node. With reference to the Sensor_ID, the information is being transmitted. The above process is represented in Algorithm 1. Table 1 clearly mention the particles fitness value in global iteration and is showed in Fig. 2. Algorithm 1. ModifiedPSO-Clustering() Procedure ModifiedPSO-Clustering() Begin Initialize population of n particles with random positions and velocities; while target fitness or maximum iteration is not attained do for each particle p in N do Calculate the fitness value (fp) of each particle (Eq. (18)); if (fp > f(pBest)) pBest = fp; end for gBest = max(pBest in P); for each particle p in N do Calculate velocity (Eq. (26)); Calculate position (Eq. (27)); end for end while End ModifiedPSO-Clustering

ð22Þ

Please cite this article in press as: Deepa, O., Suguna, J. An optimized QoS-based clustering with multipath routing protocol for Wireless Sensor Networks. Journal of King Saud University – Computer and Information Sciences (2017), https://doi.org/10.1016/j.jksuci.2017.11.007

7

O. Deepa, J. Suguna / Journal of King Saud University – Computer and Information Sciences xxx (2017) xxx–xxx Table 1 Sample Particles Fitness value in the Global best iteration. Sensor-ID

7

42

38

19

40

23

14

33

28

Fitness value

15

13

12

11

10

9

7

6

5

The LinkPerformance algorithm takes QoS metrics fRlink ; Elink ; Dlink g in order to calculate the cost of link between itself and its direct neighbor. The algorithm proficiently avoids for forming loops/cycles by using hop count and distance from each node to reach sink, which avoid deadlock problem in the network. Algorithm 2. Link Performance() Procedure LinkPerformance (Edgevw) Begin Initialize the parameters; Rvw = Psink/Psource; Calculate ET(n,d)and ER(n); Evw = ET(n,d) + ER(n); Calculate dtrans, dpropa, dprocess, dqueue; Dvw = dtrans + dpropa + dprocess + dqueue; if (Rvw >= Rreq && Evw <= Emin && Dvw<= Dreq) costvw = Rreq/Rvw + Evw/Emin + Dvw/Dreq; return (costvw) End

Fig. 2. Cluster Formation for WSN in the Sink coverage area.

Afterwards, to form a cluster, cluster head broadcasts the advertisement packet in its coverage area. Some sensors receive more than one advertisement and will select the cluster head based on higher Received Signal Strength Indication (RSSI) and reply the joining request packet. After receiving all the joining request packets, the cluster head sends the member information to the sink. For reducing the congestion, the cluster head generates the time-slot schedule for its members based on TDMA (Cionca et al., 2008) and sends to the cluster members, which is used for the collision-free communication between the cluster member and the cluster head.

4.2. Route discovery phase In this phase, each sensor broadcasts an application-specific threshold values fRreq ; Emin ; Dreq gto all its active neighbor sensors through route-request message. The neighbor sensors check whether it satisfies the application-specific threshold values and then it sends route-reply message. The format of route-request and route-reply message is shown in Fig. 3.

Hence, each sensor must also store QoS constrains satisfied neighbor table (Table 2) for link values associated with each neighbor and updated in order to select the next node with minimum cost. Thus, the OQoS-CMRP finds a set of available paths, P ¼ fpath1 ; path2 ; . . . pathnp g from the sink to all sensors by satisfying the following objective function

f : maxðf R Þ; minðf E Þ; minðf D Þ

ð28Þ

which can be written as np X fp  W

ð29Þ

p¼1

where W is the weight set of non-negative QoS threshold values required by an application. In Eq. (28), the first term specifies the maximum data transmission reliability, the next term denotes the minimum energy consumption and the last term specifies the minimum amount of delay in data transmission. Hence, the objective function is to minimize the data transmission delay while maximizing the data transmission reliability. And also it minimizes data transmission power in order to extend the network lifespan. These QoS constrains are subject to

f R P Rreq

ð30Þ

X min f E

ð31Þ

f D 6 Dreq

ð32Þ

Then, by using these constrains, the cost of link value ðcostv w Þ is calculated by

costv w ¼

Rreq Ev w Dv w þ þ Rv w Emin Dreq

ð33Þ

Fig. 3. Format of Route-request and Route-reply Message.

Table 2 QoS Constrains Satisfied Neighbor Table for node (v). Neighbor ID

1

2

.

.

n

Rlink Elink Dlink Hop count Distance to Sink Cost

Rv 1 Ev 1 Dv 1 hopv 1 d1;sink costv 1

Rv 2 Ev 2 Dv 2 hopv 2 d2;sink costv 2

. . . . . .

. . . . . .

Rv n Ev n Dv n hopv n dn;sink cost v n

Please cite this article in press as: Deepa, O., Suguna, J. An optimized QoS-based clustering with multipath routing protocol for Wireless Sensor Networks. Journal of King Saud University – Computer and Information Sciences (2017), https://doi.org/10.1016/j.jksuci.2017.11.007

8

O. Deepa, J. Suguna / Journal of King Saud University – Computer and Information Sciences xxx (2017) xxx–xxx

The sink is responsible for establishing a communication path to all sensors by executing greedy algorithm SingleSinkAllDestination() to find optimal solution from the set of feasible solutions. First, the sink starts to select all cluster head in the coverage area. All cluster head’s forward route-request message to its coverage area and find the next hop neighbor node with minimum cost. The neighbor node is added to the shortest path list. Again, the selected neighbor node find the next hop neighbor node with minimum cost. This local search procedure is introduced in OQoS-CMRP which refines the optimal solution at the end of each iteration. This process is repeated until the shortest path to all active sensors in the network has been found. Algorithm 3. SingleSinkAllDestination() Procedure SingleSinkAllDestination() Begin Initially all vertices visited array is set to 0 i = 0; while (i <= n) if (visited[i] = 0) then path_cost = hop_count = 0; Rpath[i] = Epath[i] = Dpath[i] = 0; call OptimalRouteDiscovery(i); i = i + 1; End Procedure OptimalRouteDiscovery (v) Begin min_cost = -1; vistited[v] = 1; for each vertex w adjacent from v do until to cover half of the vertices in the coverage area near by the sink if (visited[w] = 0) then /⁄Choose the best node among their neighbor nodes by calculating cost of link performance based on QoS metric {Rlink, Elink, Dlink}, the distance from sink and number of hop counts ⁄/ costvw = call LinkPerformance (Edgevw); /⁄ Compare it with all adjacent nodes in the half of coverage area, then choose the best node⁄/ if (costvw
Fig. 4. Data transmission in OQoS-CMRP.

data can be defined as the transmission strategy (TS) and is given by



TS ¼

i

if T P i P 2; multipath routing

1 single path routing

ð34Þ

where T is an application-specific threshold value and selection of pathi ; inp. In the Round-robinPathsSelection algorithm, the selection of next path is limited to a subset of the set of best-case paths based on the T value, which is based on select the best-case path or paths from subset of paths. To distribute the traffic load among two or more paths at fixed time slice by using the concept of roundrobin scheduling algorithm. When the data reaches the sink, an acknowledgment packet transmits back to the cluster head. If the cluster head does not receive the acknowledgment from the sink, it retransmits the data. Frequently, the sink monitors the residual energy of sensor and information about network topology. If sink finds sensor residual energy below the threshold value, it selects another available path from the subset. Algorithm 4. Round-robin Paths Selection for data transmission Procedure Round-robinPathsSelection() Begin Initialize the parameters sort Ascending order by path_cost; /⁄ when an event occurred⁄/ while (!end of event) if (path_count = 1) single path routing; else Select any one of the path from 1 to T by using round robin method and then route data to sink; Re2e = 1-(Re2e X (1-Rpath[i])); Ee2e = Ee2e + Epath[i]; De2e = De2e + Dpath[i]; End while End

4.3. Data transmission phase The proposed algorithm uses both single path and multipath routing in order to select the near optimal route with minimum cost, hop count and maximum residual energy. In Fig. 4, when an event occurs, source sensor sends data towards its shortest path in the fixed time slice period. Another time slice, the next data packet will also be sent to an alternative neighbor in its coverage area, which is the best-case multipath routes to reach the sink and satisfies QoS constrains. The number of paths used to route

4.4. Rerouting or Re-clustering At the regular interval of time, the sink initiated the process of rerouting and re-clustering. The sink monitors the residual energy of each sensor in the network. If any sensor falls below the threshold value, it initiates the process of rerouting and re-clustering based upon the role of the sensor. If the sensor is the relay node

Please cite this article in press as: Deepa, O., Suguna, J. An optimized QoS-based clustering with multipath routing protocol for Wireless Sensor Networks. Journal of King Saud University – Computer and Information Sciences (2017), https://doi.org/10.1016/j.jksuci.2017.11.007

9

O. Deepa, J. Suguna / Journal of King Saud University – Computer and Information Sciences xxx (2017) xxx–xxx

5. Results and discussion 5.1. Simulation results Initially, some application-specific QoS threshold metric values are declared and established for a system model. The objective function of OQoS-CMRP for WSN is to form cluster, select cluster head and to find the shortest path between the sink and all nodes in the network. The algorithm takes node forwarding capability, total length of route, queue length and node residual energy. In this section, the simulation results are presented using Network Simulator tool, version 2.35 (www://http.isi.edu/nsnam/ns/). The experiments are performed with 50 to 250 nodes which are randomly placed in rectangular fields, 300m  200m for different simulation. A rectangular field is chosen so that the transmission for far away nodes can also be evaluated. IEEE 802.15.4 is the de facto standard that defines the physical and MAC sub-layer for accessing the communication medium in the low rate personal area networks (Lee, 2006). The data packet size is fixed at 128 bytes. Each sensor is assumed to have an initial energy of 3 joules. The network is defined to have 250kps bandwidth. The parameters and their specifications chosen are shown in Table 3. 5.2. Performance analysis

EE-LEACH Network Lifetime(s)

of any path, the sink selects another available path to exclude that node. If the sensor is the cluster head, the sink selects another cluster head and corresponding path. This process increases the network lifetime.

EPSO-CEO

OQoS-CMRP

25000

20000

15000

10000

50

100

150

200

250

Number of nodes Fig. 5. Network Lifetime.

Fig. 5 shows the network lifetime comparison between proposed OQoS-CMRP and existing EE-LEACH, EPSO-CEO routing protocols. It is found that the proposed OQoS-CMRP enhance the lifetime of network because of clustering with multipath routing. 5.2.2. Total energy consumption The energy consumption of a sensor is based on sensing, computing and communication. The energy consumption for a node is calculated by

Ec ¼ Ei  Er

5.2.1. Network lifetime Sensor network lifetime depends on the number of active sensors and connectivity among them in the network, so energy must be used efficiently in order to maximize the network lifespan.

Table 3 NS2 Simulation Configuration Parameters. Parameters

Specifications

Radio Propagation Model MAC Type Antenna Type Simulation area Link bandwidth / Data rate Radio Frequency Number of nodes Radius (r) Simulation time Channel Type Queue Type Data Packet size Control Packet size Initial energy Transmission power consumption Reception power consumption Eelec efs emp do Acceleration coefficient a1,a2 Routing Protocol Traffic source

Two-ray ground reflection model IEEE 802.15.4 Omnidirectional 300X200m 250kbps 2.4 GHz 50, 100, 150, 200, 250 40 m 500 s Channel/wireless Priority Queue 128 bytes 32 bytes 3J 0.002*dist J 0.02 J 50 nJ/bit 1011 J/bit/m2 1.3  1015 J/bit/m4 87 m 0.5 OQoS-CMRP CBR

ð35Þ

where Ec is energy consumed by a sensor, Ei is an initial energy of the sensor and Er is the remaining energy available in the sensor. The total energy consumption is calculated by adding total amount of energy consumed by all nodes to transmission and reception of control and data packets over the simulation time. This parameter is vital as it determines the overall network lifetime. Fig. 6 shows that proposed OQoS-CMRP effectively consumes minimum amount of energy because of the less number of failure paths and optimal selection of multipath compared to the other routing protocols. 5.2.3. Average residual energy It is the average of total residual energy of all sensors over the simulation time.

EE-LEACH Total Energy Consumption (J)

The proposed algorithm OQoS-CMRP for WSN is compared with other existing routing protocols namely EE-LEACH and EPSO-CEO using the same initial values and the same scenario with the different parameters. The following are the performance of QoS metrics used in our simulations.

EPSO-CEO

OQoS-CMRP

20 15 10 5 0

50

100

150

200

250

Number of nodes

Fig. 6. Total Energy Consumption.

Please cite this article in press as: Deepa, O., Suguna, J. An optimized QoS-based clustering with multipath routing protocol for Wireless Sensor Networks. Journal of King Saud University – Computer and Information Sciences (2017), https://doi.org/10.1016/j.jksuci.2017.11.007

10

O. Deepa, J. Suguna / Journal of King Saud University – Computer and Information Sciences xxx (2017) xxx–xxx

EPSO-CEO

OQoS-CMRP

EE-LEACH

3

EPSO-CEO

OQoS-CMRP

100 90 PDR (%)

Average Residual Energy (J)

EE-LEACH

2.9

80 70 60

2.8

50 50

100

150

200

100

250

200

300

400

500

Simulation Time (s)

Number of nodes

Fig. 7. Average Residual Energy.

Fig. 9. Packet Delivery Ratio.

From Fig. 7, it is observed that the average residual energy of proposed OQoS-CMRP protocol is higher when compared with other competitive protocols.

Fig. 10 despites the impact of the number of nodes to the normalized overhead added to the WSNs. The increasing number of nodes causes more normalized overhead as expected. It concludes that the proposed OQoS-CMRP have significantly lower amount of normalized overhead than other routing protocols.

5.2.5. Packet Delivery Ratio (PDR) PDR shows the data transmission reliability. PDR is the ratio of actual packets successfully received by the sink to the total packets sent by the source in a network. Fig. 9, clearly shows that the PDR of proposed OQoS-CMRP protocol is 91.05% which is 10.63% and 6.55% higher than EE-LEACH and EPSO-CEO routing protocols respectively.

EE-LEACH

10 5 0 50

EPSO-CEO

EE-LEACH End-to-End Delay (ms)

Throughput (Kbps)

35

100

200

300

400

Simulation Time (s)

Fig. 8. Throughput.

500

150

200

250

Fig. 10. Normalized Overhead.

65

45

100

Number of nodes

OQoS-CMRP

55

OQoS-CMRP

15

5.2.6. Normalized Overhead The normalized overhead is defined as the total number of control packets normalized by the total number of received data packets.

EE-LEACH

EPSO-CEO

20 Normalized Overhead (%)

5.2.4. Throughput It is defined as the number of packets received at a particular point of time. Fig. 8 shows the results of the proposed OQoS-CMRP which has high throughput because of multipath routing. The proposed method achieves throughput value of 55.886Kbps which is 9.292Kbps and 6.4Kbps higher than EE-LEACH and EPSO-CEO respectively.

EPSO-CEO

OQoS-CMRP

0.06 0.05 0.04 0.03 0.02 0.01 0

100

200

300

400

500

Simulation Time (s) Fig. 11. End-to-End Delay.

Please cite this article in press as: Deepa, O., Suguna, J. An optimized QoS-based clustering with multipath routing protocol for Wireless Sensor Networks. Journal of King Saud University – Computer and Information Sciences (2017), https://doi.org/10.1016/j.jksuci.2017.11.007

O. Deepa, J. Suguna / Journal of King Saud University – Computer and Information Sciences xxx (2017) xxx–xxx

5.2.7. End-to-End delay The end-to-end delay is measured as the total amount of time taken by a packet to be transmitted across a network from source to sink. It considers all types of delay such as transmission delay, queuing delay, processing delay and so on. This metric indicates the robustness of the routing protocol. The impact of proposed OQoS-CMRP on end-to-end delay is shown in Fig. 11, which is minimum when compared with existing competitive protocols. Simulation results clearly demonstrate that the proposed OQoSCMRP for WSN is efficient and also verified that the protocol is practicable. 6. Conclusion and future work Proposes an Optimized QoS based Clustering with Multipath Routing Protocol (OQoS-CMRP) for WSNs is to establish near optimal routes for data transmission under the multi-constrained QoS. This research work focuses on energy conservation in each sensor of sink vicinity by solving energy hole problem using Modified PSO-based clustering algorithm and to find the next hop neighbor based on SingleSink-AllDestination algorithm. Furthermore, OQoSCMRP achieves better load balancing by dynamically choosing alternate path from subset of best-case paths to transmit data. The simulation results are evaluated and compared with the existing protocols EE-LEACH and EPSO-CEO. The performances of these protocols are evaluated with respect to the metrics such as network lifetime, total energy consumption, average residual energy, throughput, packet delivery ratio, normalized overhead and endto-end delay. The simulation results proved that the proposed protocol achieves better communication reliability with minimum delay while maintaining reasonable energy consumption and enhance lifetime of network. Hence, the principle of optimality holds. In future, the proposed protocol can be employed for enhancing the efficiency of various applications like multimedia messaging, video calling, etc. The proposed protocol mostly deal with energy efficiency in routing protocol can be further extended to improving energy efficiency in MAC layer. Also the concepts of enhancing fault tolerance, security and mobility in communication will be performed. References Akyildiz, I.F., Su, W., Sankarasubramaniam, Y., Cayirci, E., 2002. Wireless sensor networks – A survey. IEEE Commun. Mag. 40 (8), 102–114. Romer, K., Mattern, F., 2004. The design space of Wireless Sensor Networks. Proc. IEEE Conf. Wireless Commun. 11 (6), 54–61. Akkaya, K., Younis,, 2005. A survey on routing protocols for wireless sensor networks. Elsevier Ad Hoc Network J. 3 (3), 323–349. Yick, Jeniffer, Mukherjee, Biswanath, Ghosal, Dipak, 2008. Wireless sensor network survey. Comput. Netw. 52, 2292–2330. Toldan, Phuntsog, Kumar, Aja Ahamd, 2013. Design Issues and various Routing Protocols for Wireless Sensor Networks ISBN: 978-93-82338-79-6. In: Proc. of National Conference on New Horizons in IT, pp. 65–67. Villalba, L.J.G., Orozco Ana, L.S., Cabrera, A.T., Abbas, C.J.B., 2007. Routing protocols in wireless sensor networks. IEEE Trans. Parallel Distrib. Syst., 919–931 Sohrab, K., Gao, J., Ailawadh, V., Pottie, G.J., 2000. Protocols for self-organization of a Wireless Sensor Network. IEEE Personal Commun. J. 7 (5), 16–27. Heinelman, W.R., Chandrakasan, A., Balakrishnan, H., 2002. Energy-Efficient Communication Protocol for Wireless Microsensor Networks. In: Proc. of the 33rd Annual Hawaii International Conference on System Sciences, USA, pp. 4–7. Rohini, Sharma, Lobiyal, D.K., 2015. Proficiency analysis of AODV, DSR and TORA Adhoc routing protocols for energy holes problem in wireless sensor networks. Procedia Comput. Sci. 57 (2015), 1057–1066. Mohemed, Reem E., Saleh, Ahmed I., Abdelrazzak, Maher, Samra, Ahmet S., 2017. Energy-efficient routing protocols for solving energy hole problem in Wireless Sensor Networks. Comput. Netw. 114 (2017), 51–66.

11

Amit, Sarkar, Senthil Murugan, T., 2016. Routing protocols for Wireless Sensor Networks: What the literature says? Alexandria Eng. J. 55 (4), 3173–3183. Ganaesan, D., Govindan, R., Shenker, S., Estrin, D., 2001. Highly-resilient, EnergyEfficient Multipath Routing in Wireless Sensor Networks. In: ACM Mobile Computer Communication Review (MC2R), pp. 11–25. De, S., Qiao, C., Wu, H., 2003. Meshed, multipath routing with selective forwarding: an efficient strategy in wireless sensor networks. Wireless Sensor Network 43, 481–497. Younis, O., Fahmy, Sonia, 2004. Distributed clustering in Ad-hoc sensor networks: a hybrid, energy-efficient approach. Proc. of IEEE INFOCOM, an extended version appeared in IEEE Transactions on Mobile Computing. Yang, Y., Bai, E., Hu, J., Wu, W., 2010. MRBCH: a multi-path routing protocol based on credible cluster heads for wireless sensor networks. Int. J. Commun., Network Syst. Sci. 3 (8), 689–696. Arumugam, G.S., Ponnuchamy, T., 2015. EE-LEACH: development of energy-efficient LEACH protocol for data gathering in WSN. EURASIP J. Wireless Commun. Netw. 2015 (1), 1–9. Bhuyan, Bhaskar et al., 2010. Quality of Service (QoS) provisions in wireless sensor networks and related challenges. Wireless Sensor Network 2, 861–868. Sohrabi, K., Gao, J., Ailawadhi, V., Pottie, G.J., 2000. Protocols for Self-organization of a Wireless Sensor Network. IEEE Pers. Commun. 7 (5), 16–27. He, T. et al., 2003. SPEED: A stateless protocol for real-time communication in sensor networks. In: Proc. of the IEEE International Conference on Distributed Computing Systems, pp. 46–55. Felemban, E., Chang-Gun, L., Ekici, E., 2006. MMSPEED: multipath multispeed protocol for QoS guarantee of reliability and timelines in Wireless Sensor Network. IEEE Trans. Mob. Comput. 5 (6), 738–754. Huang, X., Fang, Y., 2008. Multi-constrained QoS multipath routing in wireless sensor networks. ACM Wireless Netw. 14 (4), 465–478. Alwan, Hind, Agarwal, Anjali, 2013. MQoSR: a multiobjective QoS routing protocol for wireless sensor networks. ISRN Sensor Netw. 2013, 1–12. Bagheri, T., Ghaffari, A., 2011. RECM: Reliable and Energy effective Clustering based Multipath routing algorithm for Wireless Sensor Networks. In: proc. of IEEE World Congress on Information and Communication Technologies, pp. 1340– 1345. Mazaheri, M.R., Homayounfar, B., Mazinani, S.M., 2012. QoS based and energy aware multipath hierarchical routing algorithms in WSNs. Wireless Sensor Netw. 4, 31–39. Almalkawi, I.T., Zapata, M.G., Al-Karaki, J.N., 2012. A cross-layer-based clustered multipath routing with QoS-aware scheduling for Wireless Multimedia Sensor Networks. Int. J. Distrib. Sensor Netw., 1–11 Sharma, Suraj, Jena, Sanjay Kumar, 2015. Cluster based multipath routing protocol for wireless sensor networks. ACM SIGCOMM Computer Commun. Rev. 45 (2), 15–20. Kennedy, J., Eberhart, R., 1995. Particle Swarm Optimization. In: Proc. of IEEE International Conference on Neural Networks, pp. 1942–1948. Del Valle, Y., Venayagamoorthy, G.K., Mohagheghi, S., Hernandez, J.C., Harley, R., 2008. Particle Swarm Optimization: Basic concepts, variants and applications in power systems. IEEE Trans. Evol. Comput. 12 (2), 171–195. Latiff, N.M.A. et al., 2007. Energy-aware clustering for Wireless Sensor Networks using Particle Swarm Optimization. In: Proc: of 18th Annual IEEE International Symposium on Personal, Indoor and Mobile Radio Communication, pp. 1–5. Singh, B. et al., 2012. A novel energy-aware cluster head selection based on Particle Swarm Optimization for Wireless Sensor Networks. Human-Centric Comput. Inform. Sci. 2 (1), 2–13. Elhabyan, Riham S., Yagoub, Mustapha C.E., 2014. PSO-HC: Particle Swarm Optimization Protocol for Hierarchical Clustering in Wireless Sensor Networks 978-1-63190-043-3. In: 10th IEEE International Conference on Collaborative Computing: Networking, Applications and Worksharing (CollaborateCom 2014), pp. 417–424. Srinivasa Rao, P.C. et al., 2016. PSO-based multiple-sink placement algorithm for protracting the lifetime of Wireless Sensor Networks. In: Proc. of the Second International Conference on Computer and Communication Technologies. Springer, pp. 605–616. Vimalarani, C., Subramanian, R., Sivanandam, S.N., 2016. An enhanced PSO-based clustering energy optimization algorithm for wireless sensor network. Sci. World J. 2016, 1–11. Deepa, O., Karthikeyani Visalakshi, N., 2016. A Self-Optimized QoS aware RED-ACO routing protocols for wireless sensor networks. Middle East J. Sci. Res. 24, 224– 230. Kulothungan, K., Ganapathy, S., Indira Gandhi, S., Yogesh, P., Kannan, A., 2011. Intelligent secured fault tolerant routing in wireless sensor networks using clustering approach. Int. J. Soft Comput. 6 (5–6), 210–215. Cionca, V., Newe, T., Dadarlat, V., 2008. TDMA protocol requirements for Wireless Sensor Networks. In: Proc. of the IEEE second International Conference on Sensor Technologies and Applications, pp. 30–35. www://http.isi.edu/nsnam/ns/. Lee, J.S., 2006. Performance evaluation of IEEE 802.15.4 for low rate Wireless Personal Area Networks. IEEE Trans. Consum. Electron. 52 (3), 742–749.

Please cite this article in press as: Deepa, O., Suguna, J. An optimized QoS-based clustering with multipath routing protocol for Wireless Sensor Networks. Journal of King Saud University – Computer and Information Sciences (2017), https://doi.org/10.1016/j.jksuci.2017.11.007

12

O. Deepa, J. Suguna / Journal of King Saud University – Computer and Information Sciences xxx (2017) xxx–xxx

O. Deepa received the M.Sc., and M.Phil., degrees in Computer Science from the Bharathidasan University, Tiruchirapalli, Tamil Nadu, India in 2000 and 2005 respectively. She qualified the Government of Tamil Nadu, State Eligibility Test for Lectureship (SET-2012) in the subject Computer Science and Application. She is currently pursuing the Ph.D. degree with the Department of Computer Science, Bharathiar University, Coimbatore. Her current research focuses on wireless communications, routing algorithm design, and performance evaluation in wireless sensor and ad hoc networks and other computational intelligence techniques as well as their applications to real-world problems.

Dr. J. Suguna received the Master’s degree in Mathematics from Annamalai University, Chidambaram in 1988 and the Ph.D. degree in Computer Science from the Bharathiar University, Coimbatore in 2009. She is currently an Associate Professor with the Department of Computer Science, Vellalar College for Women (Autonomous), Erode, Tamil Nadu. Her research interests are AI, Data Mining, Text Mining and Image Processing. She is the author or co-author of over 30 publications in journals, conference proceedings and book chapters. She has presented a paper in an International Conference held at Cincinnati University, Cincinati, Ohio, USA. She has produced over 17 M.Phil Scholars in Computer Science and guiding 8 Ph.D. Scholars at present.

Please cite this article in press as: Deepa, O., Suguna, J. An optimized QoS-based clustering with multipath routing protocol for Wireless Sensor Networks. Journal of King Saud University – Computer and Information Sciences (2017), https://doi.org/10.1016/j.jksuci.2017.11.007