统计优先级多址接入协议的改进与仿真

统计优先级的多址接入（Statistical Priority-based Multiple Access，SPMA）协议的优良性能符合未来数据链的发展趋势。经典的SPMA协议中存在固定门限设置导致的吞吐量下降、退避时间设置过于简单和低优先级分组“饥饿”等三个问题。针对这些问题，提出了统计差值退避算法和虚拟时间戳排队算法。统计差值退避算法使用负载统计量和优先级门限的差值，计算出合理的分组退避时间。虚拟时间戳排队算法根据接入带宽的最低要求，按照虚拟完成时间大小进行分组接入。结合这两个算法，设计了基于SPMA的介质访问控制层改进协议，并使用OPNET软件对改进协议进行了仿真。仿真结果表明，相比于SPMA协议的传统算法，改进协议的系统吞吐量更大且更稳定，优先级平均时延更低，同时保证了低优先级业务的最小带宽接入需求。

Improvement and simulation of statistical priority-based multiple access(SPMA) protocol

The excellent performance of Statistical Priority-based Multiple Access(SPMA) protocol is in line with the development trend of future data link.However,there exist three problems in the classical SPMA protocol,such as throughput degradation caused by fixed threshold setting,simple back-off time setting and low priority packet starvation.To solve these problems,statistical difference back-off algorithm and virtual timestamp queuing algorithm are proposed.Statistical difference back-off algorithm uses the difference between load statistics and priority threshold to calculate the reasonable back-off time.According to the minimum requirement of access bandwidth,the virtual timestamp queuing algorithm performs packet access according to the virtual completion time.Combined with the above two algorithms,an improved medium access control layer protocol based on SPMA is designed and simulated by OPNET.The simulation results show that compared with those of the traditional algorithm of SPMA protocol,the system throughput of the improved protocol is larger and more stable,the average priority delay is lower,and the minimum bandwidth access requirements of low priority services are guaranteed.