Sunteți pe pagina 1din 7

International Journal of Computational Intelligence and Information Security, October 2014 Vol. 5, No. 7 ISSN: 1837-7823

Multi-QoS Routing in MANET : A Survey

S. Beski Prabaharan and Dr. R. Ponnusamy Research Scholar, Bharathiar University, Coimbatore.

Principal, Rajiv Gandhi College of Engineering, Sriperumpudur, Chennai.


Efficient routing in a Mobile Ad-hoc Network still proves to be a challenge even with the availability of various routing techniques. Each of these algorithms has their own pros and cons. Each of them deal with a specified scenario, providing certain enhancements to the users regular systems, such as security, anonymity, faster transmission, low jitter and delay etc. This paper discusses the working of MANETs, the node categories and provides a survey on the available routing strategies in MANET, along with their pros and cons. This identifies the necessary components for a MANET routing algorithm. The necessity of a Multi-QoS and multipath routing in MANET is discussed and challenges involved in such a scenario are conversed. The concept of evolutionary algorithms and their role in MANET routing, along with the advantages they have over conventional methods are put forth. Usage of evolutionary algorithms in a MANET scenario and areas of improvement are instigated. Future research directions provide various branches in the area of MANET routing where improvements can be made.

Keywords :MANET routing; Evolutionary Algorithms; Multi-Qos Routing; Multipath routing

1. Introduction

A mobile ad hoc network (MANET) is a self-configuring infrastructure-less network of mobile devices connected wirelessly to each other. It does not have a centralized administration, and each device in a MANET is free to move independently in any direction, and will therefore change its links to other devices frequently. They are self-configuring, self-organizing and self-maintaining structures, hence they constitute a dynamic structure. The most important constraint in a MANET is that they have limited resources at their disposal; hence every node in a MANET needs to be aware of their resource status, in order to sustain in the network. Due to their wireless nature, direct communication might not be always possible, hence they communicate in a multi-hop fashion, which apparently means that each node must forward traffic unrelated to its own use, and therefore be a router. Further, the availability of every node is equally important in order to stabilize the overall performance of the network. The primary challenge in building a MANET is equipping each device to continuously maintain the information required to properly route traffic. The existence of an effective routing protocol becomes mandatory to ensure successful transmission of messages. The major functionality of a routing protocol is to update the network topology to ensure up to date node locations, and to find the optimal route from a source node to its destination. The routing protocols can be broadly classified into two categories; proactive and reactive. The proactive protocols (DSDV) initially build their routing tables and periodically exchange routing information among them. This generates a large number of control messages in the network, and using previously generated routes in a dynamic environment might prove to be futile. To overcome the disadvantages of the proactive routing protocols, reactive routing protocols have been proposed. They discover routes only when there is a necessity, hence can cope up with the dynamic environment. Further, due to the distributed and ad-hoc property of MANETs, node movement cannot be restricted. Further, nodes also tend to move in different velocities. This stochastic nature of the network nodes makes it more difficult to define a routing protocol that operates in such a dynamic environment. Jun Li et. al., describes the impact of swarm mobility on routing protocols in [24]. It proposes the Markov’s swarm mobility model that provides a probability distribution function for modeling time-dependent changes in the network.

2. Manet Node Categories And Their Working

Nodes in a MANET can be divided into two categories; Selfish and Altruistic [14]. The selfish nodes lack co- operation during the packet forwarding mechanism. These nodes tend to drop the packets that are passed through them. The behavior of a selfish node is in such a manner that is against the basic working of a MANET, i.e. node co-operation


International Journal of Computational Intelligence and Information Security, October 2014 Vol. 5, No. 7 ISSN: 1837-7823

during the process of message transmission. Legacy routing mechanisms usually isolate these nodes from regular transmissions. Altruistic nodes are the ones that co-operate with other nodes (transmit other node’s packets) during transmission. Altruism is usually rewarded with incentives. Even though selfishness in nodes seems to be a drawback that should be eliminated from a healthy network scenario, both the properties play a vital role in the sustainability of a network. A node remaining altruistic will have a higher rate of depletion. In order for a network to remain active, all nodes in a network should exist in the network. But due to routing mechanisms, some nodes are more utilized for transmission than the others, hence the power depletion rates in some nodes are much higher, which eventually leads to reduced power. A node starts behaving selfish only when it contains very less power at its disposal, and forwarding the packets of other nodes will lead to further power depletion and eventually a dead node. A dead node in a network will reduce the connectivity structure of a network, hence the node chooses to transform itself into a selfish node. Even though this problem can be solved by providing equal probability of transmission to all the nodes, this mechanism will not work in an efficient manner. All nodes remain selfish or altruistic at some point in the network depending on the average power present in all the network nodes. A tradeoff exists when a node chooses to act selfish or altruistic based on the global network welfare. Selfishness / altruistic nature here is not just about availability, because it will affect almost all other QoS constraints. A multi-criteria QoS based equilibrium must be achieved. Being selfish could compromise the overall network QoS and on the other hand being altruistic could drain that particular node’s energy and could render it useless in future. A routing algorithm must consider the pros and cons and must try to achieve an equilibrium based on requirements. Altruistic nodes consider social welfare to be important and hence the view is total. Obtaining a NASH equilibrium in this scenario will help us bring in stability and hence a proper routing process could be devised. Obtaining NASH Equilibrium based on Multiple QoS criteria is by itself difficult, as most of the properties are dependent on one another. Mostly in game theory the number of participants is predefined and hence an equilibrium could be easily defined but in this case the nodes could dynamically change their nature based on the prevailing network conditions. We can consider the average energy rate of the devices and those that are above or near that average could be used for altruistic routing and those far below it could be made selfish.

3. Manet Routing: A Study

A trust based routing model is proposed in [13]. The trust degree between nodes is computed with two values; direct

and indirect trust values. Account link delay is taken as the only QoS constraint. Every node tends to calculate a trust

value of each of its neighbors based on the packet transmission behavior. The trust values range from 0 to 1. Indirect trust degree is calculated from neighbor’s recommendations. This method tends to boost co-operation among nodes along with an efficient malicious node discovery. The downside of this approach is that, it suffers from slanderer and collusion attacks. i.e. a malicious node can send false recommendations encouraging packet transmission through another malicious node. Mobility aware, GPS aware and energy aware routing protocol is presented in [15]. It defines a proactive routing protocol, named MQ-Routing, which is a modified form of the Q-Routing algorithm. It helps maximize the node’s lifetime and also exhibits rapid adaptation towards the network topology changes. It includes two new metrics to the original Q-Routing methodology; the path availability and the energy. The route update rule and the exploration phases are performed in a proactive manner. The advantage of this protocol over its predecessor is that it provides a reduction in the number of protocol control messages, and proves to be an adaptable protocol under various scenarios.

A multipath link state routing protocol, the MultiPath OLSR is presented in [17]. This method is based on the OLSR

protocol. It proposes to provide flexibility and extensibility by using various link metrics and cost functions. This protocol is designed to be hybrid multipath routing protocol that combines both proactive and reactive features. It performs a periodic transmission of hello messages to discover the network topology, while the route computation is performed only during the transmission times. The core functionality of the MP-OLSR is performed in two phases; the topology sensing and route computation. Multipath Dijkstra’s algorithm is used for computation of routes. As an additional functionality, this algorithm also performs loop detection.

A QoS aware routing mechanism that concentrates on efficient routing of real-time multimedia traffic is proposed in

[19]. The network topology and bandwidth are updated based on the channel usage information embedded in the control

messages. Periodic information update is performed, and this information serves as the base for routing. When the need for a path arises, a logical path with the maximum available bandwidth that satisfies the QoS requirement of the information to be transmitted is determined and used.

A mobility based multicast algorithm is proposed in [20], that uses the mobility quotient of a node as the decision

making criteria during routing. In legacy routing mechanisms, mobility characteristics were assumed to be constant, but in the real time scenario, the mobility of a node rate of a node is stochastic. Hence this method calculated the relative mobility of nodes. Nodes with higher rate of mobility are considered to be non deterministic, hence it is always

recommended to choose nodes with lower relative mobility rates. Multicast algorithm proposed in [20] provides a


International Journal of Computational Intelligence and Information Security, October 2014 Vol. 5, No. 7 ISSN: 1837-7823

method that realistically predicts the relative mobility of each host in a periodic manner. The network is considered as a weighted graph, and minimum Steiner connected dominating set problem is used for routing, where the node’s mobility rate is considered to be the weight of the graph. A distributed learning automata based algorithm is then used to solve the problem. A stability based density adaptive routing (SDR) protocol is presented in [22]. Depending on the density of the network, various tactics are followed depending on the density of vicinity. Depending on the density, it chooses either the dense mode (D-mode) or the sparse mode (S-mode). This helps reduce overhead in a dense network and guarantee routing quality in sparse networks. SDR is a distributed routing protocol whose forwarding node determination is performed in independent and anonymous manner. The forwarding nodes are always self-selected. Relative stability of a node is used as the most important quality constraint while building a path. Nodes with higher relative stability rates have higher probability of getting selected. Further, SDR does not require routing table refresh or flooding, hence produces a very low overhead. The knowledge base of a node consists of only the local neighborhood information rather than the global network information. An anonymity and security based protocol (TARo) is presented in [23]. Anonymity is achieved using the keyed hash chain technique, and security by using one-to-many Diffie-Hellman mechanism. TARo is a fully distributed routing methodology, that does not require either trusted third party security mechanism or prior knowledge of the network topology to perform secure routing. It exhibits an on-demand routing process that sets up multiple anonymous virtual connections to the destination. Every node uses a short term pseudonym to define itself while defining the route. The onion routing mechanism is used during the route discovery phase. The message being passed is both end-to-end encrypted and hop-by-hop encrypted. The end-to-end encryption provides security to the packets, while hop-by-hop encryption achieves unlinkability.

4. Multi-Qos And Multipath Routing In Manet: Pros, Cons And Challenges

Studies conducted in [1][2] and [3] reveals that shortest path algorithms are not a good choice for MANETs. The first reason being, usage of the center of a network to carry out most of the transmissions, when compared to the other areas. This in turn leads to traffic congestion in the center of the network, which affects the networks performance to a large extent. This in turn leads to the depletion of battery in the center, while a very minimal level being utilized in the network edges. The next reason is that wireless medium is inherently unreliable, and is subject to link errors. Hence usage of a single path routing algorithm will lead to a higher probability of failed packets and retransmission. A multipath routing strategy would solve these drawbacks and provide a reliable system with less retransmission rate. Multipath routing strategies in MANET can be classified as, delay aware, reliable, overhead limited, energy efficient and hybrid. The first four strategies concentrate on a single problem, while the hybrid approach concentrates on a combination of these problems. Various design challenges exist in designing a multipath routing protocol. The initial challenge is how to discover multiple paths in a network. The reason for this being a challenge is due to the constraint that the multiple paths should be either node-disjointed or link-disjointed. Hence, as we increase the number of paths to be generated, the complexity of finding such a path increases. The second challenge is to choose the required number of paths from the available path set. Random selection of paths from the set of available paths will not be efficient. Imposing an appropriate constraints during the path selection process would provide better results. The third challenge is selecting the appropriate constraint or constraints for the process of path selection. The constraint selection process is guided by two factors. The initial factor signifying efficient transmission by providing the necessary quality constraints, while the other signifies the lifetime of the system by selecting or not selecting certain nodes depending on their current power(battery) level. The final challenge is the appropriate distribution of load in the selected paths. Selecting the same path repeatedly will lead to charge depletion in the area, hence load distribution should be in such a manner that all the paths are chosen equally without any prioritization. A round robin methodology of path selection will improve the network stability. This method though efficient, has many drawbacks. The most basic drawback being longer path generation. Even though the shortest and the optimal paths are generated, not all route requests can be served with these paths. The other available paths are also chosen with the same probability. Hence usage of longer paths cannot always be avoided. It also requires the transmission of control messages to discover certain paths, and further, it has the probability of generating a route request storm. Further, it has a very high probability of generating many duplicate packets, which leads to additional processing.

5. Evolutionary Algorithms And Their Role In Manet Routing

The evolutionary algorithms are a class population based meta-heuristic techniques that are inspired by the biological evolution of the organisms. Here, the individuals are represented by the candidate solutions, while their quality


International Journal of Computational Intelligence and Information Security, October 2014 Vol. 5, No. 7 ISSN: 1837-7823

is determined using the fitness function. The fitness function determines which organism passes on to the next

generation. The individuals are modified by repeated application of the operators. Our problem in hand, the QoS based Multipath MANET routing does not require the shortest path every time a request is made, generation of an appropriate path for transmission of data at the stipulated time, and the one that meets the QoS constraints of the application transmitting the data is satisfactory. The flexible nature of the evolutionary algorithms to adapt to the changing environment becomes an added advantage when it comes to MANETs, whose nodes are dynamically changing. Further, an optimal or near optimal routing solution would suffice. Since an evolutionary class of algorithm meets most of the requirements posed by MANETs they become the best methods for route determination rather than the classic routing algorithms that only consider the distance of transmission as its only constraint. Due to the basic nature of the problem (path finding), any heuristic or meta heuristic method that performs graph processing will prove to be a good choice. To be more specific, Swarm Intelligence based methods that help to solve combinatorial optimization problems can be used for routing based problems. The basic constraints that are to be considered for the performance analysis are discussed below:

Complexity of the fitness function

This represents the number of comparisons to be handled in-order to select a node or to perform a node’s transition from one generation to the next. Even though this does not seem to involve a lot of complexity, the complexity range

varies from a very few computations (2 in case of Simulated Annealing) to exponential (Genetic Algorithm). This determines the ease at which a candidate solution can traverse from one generation to next.

Scalability with respect to complexity

Scalability refers to how an increase in the number of candidate solutions affects the computations. While some

algorithms show resistance towards changes in the number of candidates, others exhibit a drastic increase in the number of computations involved. This plays a vital role in the number of processing elements involved during computation.

Ability to define the stop criterion/ completeness

This defines the level at which a user can determine when a computation should stop. Due to the implicit property of

the methods to determine near optimal solutions, the user should be able to define the error threshold, which defines the completeness. Further, a stopping criteria can also comprise of the maximum allowed computation time, in case of delay sensitive applications.

Getting struck in local optima

Local search techniques or techniques defined with low visibility rates have a higher probability of getting struck in

the local optima. It happen when the neighboring states do not contain a better solution or if the better solution is far away (not visible) from the current state.

Handling dynamic datasets

This defines the ability of a system to handle a dynamic scenario, where the candidate solutions are generated in a

dynamic manner. Only a few algorithms that tend to provide divergence in their solutions are capable of handling such a scenario.

Parallelization capacity

This defines the ability of an algorithm to allow parallelization of its operations (task or data parallelization). Only an algorithm that contains disjoint independent operations is capable of incorporating parallelization.

Stochastic nature

This property defines the nature of an algorithm to incorporate dynamism in its data. A sequential algorithm will

contain the least stochastic nature, while an algorithm that depends partially on the previous inputs or does not depend on the feedback will be able to handle stochastic scenarios.


This property defines the ability of a system to determine the level of comfort and the increased accuracy levels achieved by an algorithm when combined with other methodologies. Some of the frequently used meta heuristic methods in the domain of routing are discussed below, along with their pros

and cons of being used in a dynamic scenario such as a MANET environment.

5.1 Ant Colony Optimization

The Ant Colony Optimization (ACO) was proposed by Marco Dorigo [4]. This method was based on the behavior of ants during their search for food. Every ant moves through a path and finally finds the food source. On its way back, it leaves a pheromone trail marking the path it has travelled as one of the available paths for reaching the food source. This can be in simpler terms a graph traversal algorithm, finding optimal routes from a source to the destination. Trail intensity in a path (amount of pheromone deposited) and Visibility of a path (inversely proportional to the distance) plays a vital role in the selection of a path. In due course, the best trail that represents the shortest path prevails and the other paths tend to fade away due to the property of evaporation. Several variants of ACO have been proposed [5], [6], [7], [8]


International Journal of Computational Intelligence and Information Security, October 2014 Vol. 5, No. 7 ISSN: 1837-7823

and [9]. These variants differ in terms of the amount of pheromone deposit and the phase during which the pheromone deposition mechanism takes place (either after obtaining a complete path, or after the selection of every node). ACO scores moderate when it comes to the complexity of the fitness functions to be used. It performs well in terms of scalability and the ability to define stopping criterion. Further, it also performs best when it comes to handling dynamic datasets and also provides higher efficiency when parallelized and is also open towards hybridization with other methods. ACO can be well adapted to solve combinatorial optimization problems. Hence this becomes a very suitable candidate for solving the routing problem in MANET.

5.2 Artificial Bee Colony Algorithm

The Artificial Bee Colony (ABC) Algorithm is a population based search algorithm that was proposed by Karaboga in 2005 [10]. It emerged from the food foraging behavior of the honey bees. It divides the bees into three groups; employed bee, onlookers and scouts. One employed bee is considered for every available food source. An employed bee visits a food source, comes back to the nest and performs a dance. The onlookers join the employed bee depending on its dance, and starts collecting food. An employed bee whose food source is depleted becomes a scout searching for food. This method on deeper inspection reveals that it is similar to ACO in terms of food foraging and functionality.

5.3 Particle Swarm Optimization

Particle Swarm Optimization (PSO) is swarm intelligence based (social behavior) model, that was initially conceived by Kennedy, Eberhart and Shi [11] and [12]. It optimizes a problem by iteratively improving a candidate solution with regard to a given quality measure. Every particle moves in a search space, each with a certain velocity and a direction. The velocity and direction are provided for the particle to move towards a better solution. Movement of every particle is determined by the knowledge of the global best and the local best solutions. PSO is parallelizable and can handle dynamic datasets to a moderate extent. Its flexibility towards hybridization is moderate, while it remains highly scalable. Due to its properties, PSO is also one of the valid candidate for the process of MANET routing.

6. Evolutionary Algorithms For Manet: A Study

A pareto-based multi-colony multi-objective ant colony optimization algorithm is presented in [16]. Unlike other approaches, this method tends to provide a set off solutions, called the pareto set. These solutions tend to contain near- optimal solutions with a good density or spread. This method conceives an island-based model for the Ant System [4]. In this method, the ant colonies communicate with each other by migrating ants, and by following a neighborhood topology, that fits into the search space. A Hybrid Bio-inspired Bee swarm Routing (HyBR) algorithm concentrating on safety applications is presented in [18]. The HyBR is based on a continuous learning process, that considers the dynamic nature of the environment into consideration while performing routing. It combines both the methods of topology routing and geographic routing. It is also a unicast and a multicast routing methodology, that transfers packets with minimum delay and high packet delivery rate. The basic working of this method is performed by partitioning the network into sub-networks for routing. When the network is found to be dense, topology based routing is performed. It is a reactive routing method, which performs three operations; beaconing, route discovery and route repair. Geography based routing phase is carried out when the network density is low. HyBR uses a meta-heuristic method to find the shortest paths between the source and the destination. An ant based routing methodology (SARA) is presented in [21], which offer a low overhead solution by optimizing the routing process. This method contains four phases; the route discovery, route maintenance, route selection and route repair. Route discovery uses the mechanism of Controlled Neighbor Broadcast (CNB). It functions similar to the Ant Routing strategy. But the difference brought about in this mechanism is in the forwarding section. Every ant broadcasts a control message to its neighbors, but only one of the neighbor nodes broadcasts this message again. Hence the traffic overhead is reduced. In the route maintenance phase, the active session paths are refreshed using only data packets, which reduce further overhead. The route repair phase uses the deep search procedure to restrict the number of nodes used for route recovery. Hence instead of discovering a new route, SARA fixes up the broken links to determine a working path. A complete path determination procedure is performed, only when the deep search procedure fails. An Ant Colony based QoS multicast algorithm (TGBACA) is proposed in [25]. It uses a routing tree to determine the multicast route that incurs minimum cost and also satisfies the quality constraints (mostly bandwidth, delay and delay jitter). In contrast to the legacy approach that is slow and complex, TGBACA optimizes the multicast tree directly. ACO is used to control the growth of the multicast tree. The Ants represented in this method does not select a single target, instead a tree containing all the target nodes is determined. An ant selects a node and adds it to a tree, and hence the tree keeps growing until all the multicast members are added. Tree pruning is then performed to remove repeated links and the final multicast tree is determined. The pheromone updates are performed only after the construction of the entire multicast tree.


International Journal of Computational Intelligence and Information Security, October 2014 Vol. 5, No. 7 ISSN: 1837-7823

7. Research Directions

Our focus for future research will be in obtaining a tradeoff between selfishness and altruistic nodes. Attaining an equilibrium based on the game theoretic principles as provided by Nash will enable us overcome many of the problems and faults inherent in MANET including Byzantine Faults, Coalitions and Asynchronous Faults as large games are found to be inherently fault tolerant to a certain extent. So a game theoretic approach becomes very well suited for this kind of a scenario where uncertainty is part of the process. A fuzzy decision making approach for determining and relaxing the selfishness/altruistic constraints based on the traffic sensitivity of the application are to be considered. The Multi-QoS routing algorithms must be validated on the basis of CAP theorem as it states very clearly that only two of the three attributes could be guaranteed at a time. Considering such a scenario in the routing process we would like to validate the applicability of CAP theorem and also the dependencies among the various other QoS parameters and their tradeoffs. We consider the option of utilizing Pareto Sets where a set of solutions rather than a single solution is found and utilized. Our approach to multi-QoS routing will be based on Hybridized Evolutionary algorithms considering multiple objectives.

8. Conclusion

On examining the MANETs, their mode of operation and their routing mechanism, several possibilities arise due to the intrinsic nature of the MANETs. The usage of Evolutionary algorithms in the process of routing can bring about highly efficient routing mechanism in such a dynamic environment. The future research directions determine the areas that can be further pondered to bring about the hidden efficiency in the process of multi-QoS multipath routing.

9. References

[1] Kumar, Anurag, D. Manjunath, and Joy Kuri. (2004). “Communication networking: an analytical approach”.


Elsevier. Bertsekas, Dimitri P., Robert G. Gallager, and Pierre Humblet., (1987), “Data networks”. Vol. 2. Englewood Cliffs,


NJ: Prentice-hall. Pham, Peter P., and Sylvie Perreau, (2003), “Performance analysis of reactive shortest path and multipath routing mechanism with load balance”. INFOCOM. Twenty-Second Annual Joint Conference of the IEEE Computer and Communications. IEEE Societies. Vol. 1.

[4] Dorigo, Marco, Vittorio Maniezzo, and Alberto Colorni, (1996), “Ant system: optimization by a colony of


cooperating agents. Systems”, Man, and Cybernetics, Part B: Cybernetics, IEEE Transactions on 26.1: 29-41. Sorin C. Negulescu, Constantin Oprean,Claudiu V. Kifor, Ilie Carabulea, (2008), “Elitist ant system for route


allocation problem”. World Scientific and Engineering Academy and Society (WSEAS) Stevens Point, Wisconsin, USA. T. Stützle et H.H. Hoos. (2000), “Max Min Ant System”. Future Generation Computer Systems, volume 16, pages



Bernd Bullnheimer Richard F. Hartl, Christine Strau, (1997), “A New Rank Based Version of the Ant System - A


Computational Study”. Working Paper No. 1. Xiao-Min Hu, Jun Zhang, Yun Li, (2008), “Orthogonal Methods Based Ant Colony Search for Solving Continuous


Optimization Problems”. Journal of Computer Science and Technology. Volume 23, Issue 1, pp 2-18. Gupta.D.K, Arora, Singh.U.K, Gupta.J.P, (2012), “Recursive Ant Colony Optimization for estimation of parameters of a function”. Recent Advances in Information Technology (RAIT), International Conference, DOI:


10.1109/RAIT.2012.6194620, pages 448 – 454. D. Karaboga, (2005), “An Idea Based On Honey Bee Swarm for Numerical Optimization”, Technical Report-TR06,

Erciyes University, Engineering Faculty, Computer Engineering Department. [11] Kennedy, J.; Eberhart, R, (1995), “Particle Swarm Optimization”. Proceedings of IEEE International Conference

on Neural Networks IV. pp. 1942–1948.doi:10.1109/ICNN.1995.488968. [12] Shi, Y.; Eberhart, R.C, (1998), “A modified particle swarm optimizer”. Proceedings of IEEE International Conference on Evolutionary Computation. pp. 69–73. [13] Wang, Bo, Xunxun Chen, and Weiling Chang, (2013), “A light-weight trust-based QoS routing algorithm for ad hoc networks”. Pervasive and Mobile Computing. [14] Cho, Jin-Hee, and Ing-Ray Chen, (2013), “On the tradeoff between altruism and selfishness in MANET trust management”. Ad Hoc Networks 11.8 (2013): 2217-2234. [15] Macone, Donato, Guido Oddi, and Antonio Pietrabissa, (2013), “MQ-Routing: Mobility-, GPS-and energy-aware routing protocol in MANETs for disaster relief scenarios”. Ad Hoc Networks 11.3: 861-878.

[16] Mora, Antonio Miguel, (2013), “Pareto-based multi-colony multi-objective ant colony optimization algorithms: an island model proposal”. Soft Computing 17.7: 1175-1207. [17] Yi, Jiazi, (2011), “Multipath optimized link state routing for mobile ad hoc networks”. Ad Hoc Networks 9.1: 28-



International Journal of Computational Intelligence and Information Security, October 2014 Vol. 5, No. 7 ISSN: 1837-7823

[18] Bitam, Salim, Abdelhamid Mellouk, and Sherali Zeadally, (2013), “HyBR: A Hybrid Bio-inspired Bee swarm Routing protocol for safety applications in Vehicular Ad hoc NETworks (VANETs)”. Journal of Systems Architecture 59.10: 953-967. [19] Kajioka, Shinsuke, (2011), “A QoS-aware routing mechanism for multi-channel multi-interface ad-hoc networks”. Ad Hoc Networks 9.5: 911-927.

[20] Akbari Torkestani, Javad, and Mohammad Reza Meybodi, (2010), “Mobility-based multicast routing algorithm for wireless mobile Ad-hoc networks: A learning automata approach”. Computer Communications 33.6: 721-735. [21] Correia, Fernando, and Teresa Vazão, (2010), “Simple ant routing algorithm strategies for a (Multipurpose) MANET model”. Ad Hoc Networks 8.8: 810-823. [22] Liu, Weijie, and Wooju Kim, (2013), “A stability-considered density-adaptive routing protocol in MANETs”. Journal of Systems Architecture 59.9: 767-775. [23] Chen, Jiefeng Terence, Roksana Boreli, and Vijay Sivaraman, (2012), “Improving the efficiency of anonymous routing for MANETs”. Computer Communications 35.5: 619-627.

Li, Jun, et al, (2012), “Swarm mobility and its impact on performance of routing protocols in MANETs”. Computer


Communications 35.6: 709-719. [25] Wang, Hua, et al, (2011), “A tree-growth based ant colony algorithm for QoS multicast routing problem”. Expert Systems with Applications 38.9: 11787-11795.