一种基于遗传算法的信道感知顺序设计

Approach of Channel Sensing Order Based on Genetic Algorithm

认知无线电中,认知用户必须不断对待选频段进行扫描以发现可用频谱用来通信。在感知能力有限的现实环境下,认知用户逐一进行信道的感知带来的问题是,感知信道越多,耗时越长,用于通信的时间就越短。所以合理设计感知顺序将是非常关键的技术。对该问题进行了全面的分析和建模,相比前人工作又引入了空闲概率和信道容量等因素,但是该问题变成了NP难问题。为了在多项式时间内求解该问题,使用了遗传算法。不仅设计了详细的算法过程,使用了精英保留等多种加速算法收敛的技术,而且重点对交叉算子进行了研究,提出了3种可行的交叉算子:单点交叉、多点交叉和编码交叉。在仿真分析中比较了遗传算法和全搜索算法的复杂度和准确度,同时以平均吞吐量和最大吞吐量为准则,对3种交叉算子进行了仿真比较,验证了编码交叉算子的相对优越性。

In cognitive radio,the cognitive user has to sense all the candidate channels to search idle channels.In practice,its sensing ability is limited,which makes the user sense the channel in one-by-one mode.However,the more channels are to sense till an idle is found,the more time is needed to spend in the sensing process,so that the less time is remainder for transmission.Therefore,sensing order is critical for the spectrum efficiency.In this paper,the optimization of sensing order problem was modeled completely with the idle probability and channel capacity considered,which is different from the previous studies.Since the modeled problem is NP-hard,the traditional optimization lacks computation efficiency.Therefore,the genetic algorithm was adopted.Different from the classical genetic algorithm,the proposed algorithm develops a unique elite reservation strategy.Moreover,three crossing operators were proposed intensively for the problem,i.e.the single-spot crossing,the multiple-spot crossing and the coding crossing.In simulation,the computation complexity and the accuracy are compared between the genetic algorithm and brute force algorithm.After obtaining the advantage of the genetic algorithm,we compared the three proposed crossing operators in terms of expected throughput and maximum throughput,and verified the superiority of the coding crossing operator.