Reseach On Transmission Technology Based On Q-learning And Network Coding In MANETs Network

In Mobile Ad hoc Networks, according to a routing criteria(such as the number of hops),there is not only one next-hop selection for the source node or relay node to the destination. The default method is taking next-hop selection randomly, but random selection may cause congestion, delay increase and waste of network resources.This paper proposes Mobile Ad hoc Networks concurrency packets transmission strategy based on Q-learning algorithm to select the reasonable next hop node. Q-learning algorithm gets the reward through executing actions and updates the corresponding Q values. Then the node selects the next hop node based on the current Q value and action selection strategy. This paper achieves the Q-learning algorithm and performance simulation on the OPNET platform. Comparing to the random selection transmission strategy on end to end time delay performance, we can find that Q-learning algorithm transmission strategy shows much better end to end on time delay performance. This paper first uses greedy action selection strategy based on the current Q value. But we find a problem through the performance analyzing: the node makes action selection only refers to the current Q value but ignores history experience, the nodes may make a bad decision. Therefore this paper proposes the greedy action selection strategy based on the average Q value. So the node can make the global action selection strategy. We can find that the average Q value greedy action selection strategy works more stable than the current Q value greedy action selection strategy on time delay.On the other hand, how to transmit data efficiently and correctly under jamming circumstance is also a problem.This paper introduces the idea of network coding-a research hotspot in recent years and proposes the end to end transmission strategy based on random linear network coding and achieves the transmission strategy by the combination of OPNET and MATLAB. This paper compares the end to end transmission strategy based on random linear network coding with the stop-and-wait ARQ protocol, we can find that random linear network coding transmission strategy is more efficient than the stop-and-wait ARQ protocol. At last, this paper researches the effect of the encoded block size on transmission performance. It comes out with the conclusion that bigger size of encoded block has a worse real-time performance but a higher valid bit rate simultaneously.Conversely, smaller size of encoded block has a better real-time performance but a lower valid bit rate.