Enhanced AntNet Protocol for Wireless Multimedia ...

43 downloads 204 Views 385KB Size Report
The field of wireless multimedia sensor networks (WMSN) attracts more and .... Enhanced AntNet has a performance advantage over AODV. The advantage.
Enhanced AntNet Protocol for Wireless Multimedia Sensor Networks Ismail Bennis1 , Ouadoudi Zytoune2 and Driss Aboutajdine1 1

LRIT, unit´e associ´ee au CNRST (URAC29), Facult´e des Sciences, Universit´e Mohammed V - Agdal, Rabat, Maroc. 2 Universit´e Ibn Tofail, K´enitra, Maroc. [email protected], [email protected], [email protected]

Abstract. The field of wireless multimedia sensor networks (WMSN) attracts more and more the research community as being an interdisciplinary field of interest. This type of network becomes a low cost, multifunctional due to advances in micro-electromechanical systems, as well as proliferation and progression of wireless communications. However, transmitting collected multimedia information must meet the QoS criteria such as delay, bandwidth, packet loss rate, etc. Many routing protocols have been developed for WMSN. Recently, the most known ones are based on meta-heuristic, that show desirable properties of being adaptive, scalable, and robust. This paper presents a new routing protocol for WMSNs based on AntNet protocol which is inspired by the stigmergydriven shortest path following behavior of biological ants. Our aims in this work is to provide as well as possible the best QoS in terms of delay and Packet Delivery Ratio(PDR). The results of simulations compared to the AODV show that our work has better delay and PDR. Keywords: WMSN, AntNet, Delay, Routing protocol, QoS.3

1

Introduction

WSNs consist of an amount of independent nodes equipped with sensing capabilities, wireless communication interfaces, limited processing and energy resources. Regarding applications of this type of network, they are characterized in general, with low bandwidth demands, and are usually delay tolerant. Recently the availability of CMOS cameras and microphones to capture multimedia content from the environment has allowed the arrival of new network type called WMSNs [1]. This novel network will not only enhance existing sensor network applications such as tracking or home automation, but they will also enable several new applications such as multimedia surveillance sensor networks, traffic avoidance[1]. When network size scales up, routing becomes more challenging and critical. Lately, biologically-inspired intelligent algorithms have been deployed to tackle this problem [2]. Using ants and other social swarms as models, software agents can be created to solve complex problems. One of the most successful swarm 3

The final publication is available at Springer via doi:10.1007/978-3-642-40148-0 31

intelligence techniques is called Ant Colony Optimization (ACO) [3]. We can say that during these last two decades, this technique have served as an important source of inspiration for the design of novel algorithms and systems [4]. The remainder of the paper is organized as follows; in section II we discuss the implementation of AntNet for WMSN and propose our optimization for this protocol, the results of extensive simulations are shown in section III. Finally section IV, concludes the paper.

2

Implementation & Enhanced AntNet for WMSN

In this work we have made several optimizations of the initial version of the protocol (Antnet). These optimizations can be classified into two categories, the first categorie’s is related to the adaptation of the protocol in the wireless and mobile environment in view that the AntNet protocole has been designed for fixed networks. Some of this modification was inspired from others works like [5] and [6]. Also we were led to make some change in the packet classes that can be divided into three different types: Data packets represent the information which must be transmitted by the sonsor node, Control packets which contains two types : Forward ants used to search the path and Backward ants used to update the routing tables, and Hello Control packets that are used to have a list of available neighbor nodes. Also each node has two data structures the first one is a routing table Tk containing triples of a destination address d, a nexthop n used to reach that destination d and a probability Pnd . This probability value Pnd stored in the routing table express the goodness of choosing the associated nexthop n to reach P the destination d. The probabilities have to verify: n∈Nk Pnd = 1 where d ∈ [1, N ] and Nk = neighbors(k). The routing table Tk is changed by incrementing (Eq 1) or decrementing (Eq 2) the probability Pnd according to the following way : Pnd ← Pnd + r ∗ (1 − Pnd ) (1) Pmd ← Pmd ∗ (1 − r), ∀m ∈ Nk , m 6= n

(2)

Where r is a reinforcement parameter in the interval [0,1]. The second categorie’s comprises all functions and methods achieved and added in order to make the protocol more robust and more desirable for WMSNs. In what follows we will describe all these functions and methods incorporated into the protocol.

1. Optimization of the path. The goal here is to minimize the number of hops of the path that will be built by the forward ant , optimization is performed by removing redundant nodes and also nodes contained in the same neighborhood and are part of the path at its creation.

2. Initialisation of pheromone. The second change we made is to modify the manner that the next hop is chosen by the forward ant. As described in [7], a stochastic decision policy is applied to select the next node to move to, but if the destination is a neighbor, there is no need to calculate this probability. So we forward the ant directly to the neighbor. 3. Reactive mechanism. The basic version of the protocol is proactive, it tries to create a multitude of paths from fictitious sources to destinations, but this anticipation is not always useful because we can have paths that are not needed for the node source, and if we go to a large topology then it would be useless to create paths that will probably never be used. Therefore we planned to make the reactive protocol to avoid these problems. Also using the reactive manner will reduce congestion phenomena because the control packet are reduced. 4. List of ancestor. When a node needs to find a path to a destination, it generates the FANT which is responsible for finding the path. So, in order to increase the chance to find the path and accelerate research, we applied a mechanism such that each node receiving a FANT seeking a path to a particular destination, generates itself a FANT to this destination. But every node that participates in research must, in the case of success, reported its findings to the source node. To do this, it needs a list of history nodes which we called ”list of ancestors”, this list is optimized and assigned to each node receiving FANT before activating the process FANT at him. 5. List of destinations. In each node, we added a table that contains the state of this node towards each others node in the topology. we assume that the node can have three states: Transmitter :the node are the source of the trafic; Intermediate : the node is part of a path from source to destination; None :the node has no interaction with topology. We added also two lists in node : the first list will contain the destinations that the node should look the way by the FANT and this node participates as the transmitter, we call this list list dest src. The second list will contain the destinations that the node should look the way by the FANT and this node can participate as intermediate, we call this list list dest inter. 6. Update pheromone. The update of the pheromone value is governed by the equations 1 and 2, or this way of doing requires some time before converging to optimal solutions, and this becomes clearer in a large-scale topology. In order to accelerate the convergence we have changed the manner of the update, this is accomplished by penalizing the neighbors of a node that lead to the same destination but in greatest number of hops.

3

Simulation and experimental results

Our aims in this section is to compare our proposition withe AODV. The comparison regards the end-to-end delay,the packet delivery ratio (PDR) and the overhead. The simulation software used is NS2. For simplicity raisons we assum that a constant reinforcements model is used: r = C, C ∈ [0, 1]; and there is no mobility in this scenario. We have created many different problems and for each one we test the both protocols, the shown results are the average of this problems. In the following subsection we describe the environment and scenario of the simulation. Each simulation scenario is presented as follows : X nodes are randomly placed in an area of 1500*300 m2 , where: X=[20;40;60;80;100;120;140;160;180;200;220;240;260;280;300] . The data traffic is generated by 10% of nodes of topology as constant bit rate (CBR) sources sending four 512-byte packet per second. Each source starts sending at a random time between 20 and 180 seconds after the start of the simulation, and keeps sending till the end, the lenght of the simulation is 300s. At each number of nodes we repeat the simulation several times, and we calculate the average of the delays, overhead and the values of PDR found.

Fig. 1. Average delay VS Number of nodes

Fig. 2. Normalized overhead VS Number of nodes

Figure 1 shows the average delay experienced when using the both routing protocols under different number of nodes. We can see that we have the nearest delay when the number of nodes is less than 180. But, after that the dealy for AODV in scenario 1 and 2 grows up clearly to reach 2 seconds, and in case of our optimisation the delay does not exceed 0.4 seconds, this can be explained by the fact that the AODV broadcast a RRequest packet, or if the nodes have a significant number of neighbors, it will generate more packet control that will increase the congestion and thereby delaying the packet data. In Figure 2 we can see also, in case of AODV, that the overhead becomes greater after 180 nodes, but for the Enhanced Antnet the overhead remains nearly constant for all situations; as explained above once the number of nodes

Fig. 3. PDR VS Number of nodes

in the neighborhood becomes greater, the traffic generated becomes greater too, which may explain the increase in the case of AODV. Figure 3 shows clearly the gap between the value of the packet delivery ratio of the both protocols, we can perceive that the PDR for the Enhanced Antnet has some fluctuation but never drops below the 80%, in contrast to the AODV where PDR reaches very low values (up to 40%) when the number of node is 300. So, all results show that we have almost the best delay, overhead and PDR, especially when the number of node increases, which is suitable for the WMSN.

4

Conclusion and Future Work

In this paper we have described an optimized AntNet routing algorithm for WMSN, our work enhances the basic AntNet protocol which is based on the most known swarm intelligence techniques ACO. Simulation results show that Enhanced AntNet has a performance advantage over AODV. The advantage exists in terms of packet delivery ratio, average end-to-end delay, and also the overhead. For future work, there is a point that we want to improve, is to introduce mobility in simulation scenario in order to have more realistic models and to compare our work with other protocol designed for WMSN.

References 1. Ian F. Akyildiz and Tommaso Melodia and Kaushik R. Chowdhury,A Survey on Wireless Multimedia Sensor Networks, Computer Networks,51,921-960,2007. 2. R. GhasemAghaei, A. Rahman, W. Gueaieb, A.E. Saddik,Ant colony based reinforcement learning algorithm for routing in wireless sensor networks, IEEE Instrumentation and Measurement Technology Conference Proceedings, 2007. 3. M. Dorigo, C. Blum, Antcolonyoptimization theory: A survey,Theoretical Computer Science,344,243 - 278, March, 2005.

4. Muhammad Saleem and Gianni A. Di Caro and Muddassar Farooq ,Swarm intelligence based routing protocol for wireless sensor networks: Survey and future directions,Information Sciences,181,4597 - 4624,2011. 5. Di Caro, Gianni and Ducatelle, Frederick and Gambardella, Luca Maria, AntHocNet: an adaptive nature-inspired algorithm for routing in mobile ad hoc networks,European Transactions on Telecommunications,16,5,443–455,2005. 6. Camilo, Tiago and Carreto, An energy-efficient ant-based routing algorithm for wireless sensor networks, Proceedings of the 5th international conference on Ant Colony Optimization and Swarm Intelligence, Springer-Verlag,49–59,2006. 7. G.A. Di Caro and M. Dorigo, AntNet: Distributed Stigmergetic Control for Communications Networks ,Journal of Artificial Intelligence Research (JAIR), 1998.