异构密集网络下基于POMDP负载感知的负载均衡算法研究

发布时间：2018-06-09 21:27

本文选题：异构密集网络 + 负载均衡　；参考：《电子与信息学报》2017年09期

【摘要】：针对异构密集网络场景中业务不确定性而引起的网络负载不均衡,该文提出一种基于部分可测马尔科夫决策过程(POMDP)负载感知的负载均衡方法。该方法首先通过对系统用户在感知周期内数据包传输数量进行观察,预测出下一周期系统可能出现的负载状态。其次根据负载感知结果动态调整小区范围扩展偏置值(DCRE),以达到优化系统整体负载均衡性的目的。最后采用启发式算法近似求解,能够快速得到次优解。仿真结果表明,该方案能在异构密集网络下提高系统负载均衡性,同时提升了系统吞吐量与系统资源利用率。
[Abstract]:In this paper, a load balancing method based on partially measurable Markov decision process and POMDP is proposed to deal with the network load imbalance caused by traffic uncertainty in heterogeneous dense network scenarios. By observing the number of packets transmitted by the users during the perceptual period, the method predicts the possible load state of the system in the next cycle. Secondly, according to the result of load perception, we dynamically adjust the extended bias value (DCREN) to optimize the overall load balance of the system. Finally, the heuristic algorithm is used to approximate the solution, and the sub-optimal solution can be obtained quickly. Simulation results show that the proposed scheme can improve system load balance and system throughput and system resource utilization in heterogeneous dense networks.
【作者单位】：重庆邮电大学移动通信技术重点实验室;
【基金】：国家863计划项目(2014AA01A701) 国家自然科学基金(61571073)~~
【分类号】：TN929.5

【相似文献】