Research On Spectrum Allocation Algorithm Based On Energy Efficiency In Cognitive Radio Senor Network

The unlicensed spectrum,which the traditional wireless sensor network(WSN)usually works on,has been occupied and crowed by the growing facilities of other wireless communication technology.It is more serious than ever before for the lack of spectrum resource.Benefiting from the mature theory and technology progress in cognitive radio(CR),it becomes a new opportunity to relieve the strain on spectrum resource.With the capability of cognitive radio,the new network which we call cognitive radio sensor network(CRSN),can realize the real-time spectrum sense and dynamic spectrum access on the idle licensed spectrum according to its communication demands.It's not only helpful in improving the utilization rate of the licensed spectrum,but also changing the situation of spectrum scarcity in CRSN.It's worth noting that the introduction of the cognitive radio technology also brings the extra sensor node energy cost,and CRSN is usually a resource-constrained network.Therefore,both from the perspective of casing the shortage of spectrum resource and the perspective of the energy conservation,finding a more suitable dynamic spectrum allocation algorithm to optimize the throughput and the energy efficiency of the network is an urgent demand in reality.To solve this problem,this paper made a research on dynamic spectrum allocation algorithm based on energy efficiency in distributed CRSN.The main work in this dissertation is as follows:1.According to the autonomy characteristic of the cognitive nodes in the distributed CRSN,this paper proposes a modified spectrum allocation algorithm with multi-agent best response Q-Learning,which is setup in a multi-agent distributed independent reinforcement learning model under a framework of timesharing-track.The algorithm aims at maximizing both the average throughput and the average energy efficiency ratio of entire network.Finally,the convergence and effectiveness of our proposed algorithm are verified by the experiments.2.After defining the utility function with the average energy efficiency ratio of a cognitive sensor node,we model and analyze the process of power allocation in CRSN combined with the Game theory.Then the existence and uniqueness of Nash equilibrium in such a power allocation game are proved respectively.On that basis,then a joint spectrum and power allocation algorithm based on energy efficiency is proposed.By using the Q-Learning and dichotomy iteration respectively,the strategy of spectrum and power allocation can get updated continuously.Finally,the experimental data illustrates that the algorithm not only has an advantage in improving the average energy efficiency ratio and the average throughput of entire network,but also reducing the channel switching times.