认知车载网中基于簇和MAB模型的信道接入算法

A novel channel access algorithm based on clusters and MAB model in cognitive vehicular network

针对高密度车流环境下认知车载网中车辆节点对认知信道的接入问题,提出了基于簇和MAB模型的clusters-UCB信道接入算法,通过簇内成员协作提高感知学习结果的准确性,提升算法学习速度,并且簇首通过clusters-UCB算法分布式地快速搜索出最佳信道,渐近地实现最大的时隙吞吐量。仿真实验表明,提出的算法相对于多用户的UCB算法和ε-greedy算法,遗憾值更低,并且趋近于对数形式的收敛速度更快,能够有效减少访问碰撞数,保证信道接入的公平性,提高时隙吞吐量。

Considering the cognitive channel access problem of vehicle nodes in cognitive vehicular networks with heavy traffic environment, a channel access algorithm called clusters-UCB which based on clusters and MAB model was proposed. The cooperation of cluster members could improve perception accuracy and enhance the learning speed. And using improved multi-user UCB algorithm, cluster heads could quickly search out the optimal channel in a distributed way, which could make the network asymptotically achieve the optimal slot throughput. Simulation results show that with respect to UCB algorithm and ε-greedy algorithm, the regret of the proposed algorithm is lower and the speed of approaching logarithmic form is faster. What＇s more, clusters-UCB can effectively reduce the number of collisions when clusters access the cognitive channels, ensuring the fairness of the channel access and achieving better slot throughput.