区分多业务的跨层优化无线网络协议头压缩算法

在网络中，鲁棒性协议头压缩(robust header compression, ROHC)算法需要压缩端和解压端的状态同步，才能成功解压ROHC数据包，但ROHC算法的双向可靠R模式和双向优化O模式需要单独的反馈信道，增加了网络成本。针对ROHC算法的单向U模式，当无线信道质量变差时，ROHC的性能会随之下降，且现有ROHC算法没有考虑到不同业务对实时性需求不同的问题，提出了一种新的U-ROHC建模方案，引入一种部分可观测马尔可夫过程(partially observable Markov processes, POMDP)跨层优化框架来自适应调整U模型下ROHC压缩端状态，以求改善ROHC的性能。针对实时性业务和非实时性业务的不同要求自适应调整ROHC参数，使实时性业务能在解压失败的情况下具有更快速恢复双端置信的能力，而非实时性业务具有更高的压缩比，以使ROCH具备支持不同业务类型的能力。仿真结果表明，可以使ROHC-U算法在无线信道恶劣的条件下依然保持较好的压缩比和吞吐量。

Cross-layer optimization protocol header compression algorithm for multi-service differentiated wireless networks

In wireless networks, ROHC algorithm needs state synchronization between the compressor and the decompressor to successfully decompress ROHC packet. However, bidirectional reliable R mode and bidirectional optimized O mode of ROHC algorithm need separate feedback channels, increasing the cost of the network. Therefore, this paper only considers the optimization in unidirectional U mode. When the wireless channel quality deteriorates, the performance of ROHC in the U mode will decrease, and the existing ROHC algorithm does not consider the real-time requirements of different services. This paper presents an adaptive protocol header compression scheme for multi-service differentiation. The scheme proposes a new U-ROHC modeling scheme. Secondly, a partially observable Markov processes (POMDP) cross-layer optimization framework is introduced to adjust ROHC compressor state under the U model adaptively to improve the performance of ROHC. ROHC optimistic parameter is adjusted according to different requirements of real-time and non-real-time services so that real-time services can recover double-end confidence faster in the case of decompression failure, while non-real-time services have a higher compression ratio. ROHC has the ability to support different business types. Simulation results show that ROHC maintains a good compression ratio and throughput under harsh conditions of wireless channels.