Research On Opportunistic Network Routing Algorithm Based On Cellular Learning Automata

In recent years,opportunistic networks are emerged on the basis of existing Mobile Ad Hoc Networks(MANETs)to deal with the frequent network partitions and they are widely applied in different kinds of scenarios,such as emergency communication,wildlife tracking,vehicular ad-hoc network.The characteristics of opportunistic networks are intermittent connectivity,long delivery delay and frequently transmission disruption.They are different from the traditional MANETs where an end to end path from a source to a destination is absent.How to route messages in opportunistic networks is a great challenge and attracts researchers' great attentions and interests.So the new style of store-carry-forward routing paradigm is used in the opportunistic networks.In order to study the dynamics and randomness in the opportunistic networks,we introduce the theory of cellular learning automata(CLA)to describe the interaction,interrelationship and collaboration among nodes to accomplish the message routing.This is achieved by defining the states and actions of objects in the opportunistic networks corresponding to the cellular states and actions in the cellular learning automata.The rules are given to adjust the action probabilities according to the feedback from random environment.So the optimized decisions can be made at current surroundings to improve the performance of routing algorithms.The main results and contributions of our work can be summarized as follows:(1)A novel congestion control strategy based on the dynamic irregular cellular multiple learning automata is proposed using the method of updating the action probability during the interaction between cellular learning automata and its neighbors.It describes the message distribution and status in the partial network according to the identical messages situation carried by nodes and their neighbors.Messages will be removed from node buffers on the basis of numerical dropping probability when the congestion phenomenon caused by the limited buffer of nodes in the opportunistic networks occurs.The simulation results show that our proposed congestion control strategy can improve message delivery ratio,reduce network overhead ratio and message delivery latency compared with the existing strategies which only take the local information of nodes and messages into account.(2)An energy balance buffer management policy is presented according to the current energy circumstance of partial network consisting of a node and its neighbors in which nodes are located.In the proposed policy,the threshold of receiving messages will be adjusted on the basis of the local rules using the theory of cellular learning automata.Only the messages in the range of the threshold can be received when the buffer is overflow.Thus the energy waste during the frequently messages ping pang exchange between stationary peers can be reduced effectively.The simulation experiments demonstrate that the proposed policy can efficiently improve message delivery ratio,reduce network overhead ratio and message delivery latency.Also the residual energy of nodes and the standard deviation of nodes' residual energy are enhanced.The lifetime of networks is extended and the connectivity of networks is guaranteed accordingly.(3)A distinctive cellular learning automata based on routing algorithm which takes the factors of each phase in the routing procedure of store-carry-forward into account is proposed.Message delivery probability is computed and messages will be dropped according to the dropping probability when congestion occurs during the store phase.Energy consumption will be balanced according to the threshold set by the node itself which is used to accept messages in the carry phase.Connection duration between nodes will be estimated to reduce the energy waste caused by fragment messages transmission during the forwarding process.The simulation results based on the ONE platform show that,the proposed routing algorithm achieves higher delivery ratio and less overhead ratio.In addition,it gains a balance of energy consumption and an enhancement of the whole network performances.(4)A novel adaptive adjustment strategy of n-Epidemic Routing algorithm is put forward in order to adjust the parameter n dynamically in the algorithm according to the energy condition of local environment composed by a node and its neighbors.It can compensate for the weakness of imbalance energy consumption among different active nodes in the networks caused by the n-Epidemic using static parameter n.The simulation results based on the ONE platform show that the proposed routing algorithm can be more suitable for the dynamic environment in the opportunistic network.Because it utilizes the remainder energy level of nodes and all their current neighbors to adjust the parameter n periodically and dynamically,the energy consumption among nodes is balanced and the message delivery ratio is enhanced.In summary,we explore the similar features of dynamic irregular cellular learning automata and the store-carry-forward routing strategy in the opportunistic networks.And we find that the characteristics of node independent movement,limited resource and intermittent connectivity in the opportunistic networks can be decipted using dynamic irregular cellular learning automata.In the paper,we propose a series of routing models and algorithms based on the cellular learning automata with the goal of improving and enriching the research methods on the routing algorithms in opportunistic networks.The effectiveness of our models and methods is demonstrated by extensive experimements.Also these propoesed algorithms are well compatible with the existing algorithms and have great theoretical significance and applicable value in opportunistic networks environments.