Network coding for multiple unicast sessions in multi-channel/interface wireless networks

作     者：Shafieinejad, Alireza Hendessi, Faramarz Fekri, Faramarz 

作者机构：Isfahan Univ Technol Dept Elect & Comp Engn Esfahan Iran Georgia Inst Technol Dept Elect & Comp Engn Atlanta GA 30332 USA 

出 版 物：《WIRELESS NETWORKS》 (无线网络)

年 卷 期：2013年第19卷第5期

页      面：891-911页

核心收录：

学科分类：0810[工学-信息与通信工程] 0808[工学-电气工程] 0809[工学-电子科学与技术（可授工学、理学学位）] 08[工学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

主　　题：Network coding Multi-channel/interface Channel assignment Unicast routing Wireless Mesh Network 

摘      要：Throughput limitation of wireless networks imposes many practical problems as a result of wireless media broadcast nature. The solutions of the problem are mainly categorized in two groups;the use of multiple orthogonal channels and network coding (NC). The networks with multiple orthogonal channels and possibly multiple interfaces can mitigate co-channel interference among nodes. However, efficient assignment of channels to the available network interfaces is a major problem for network designers. Existing heuristic and theoretical work unanimously focused on joint design of channel assignment with the conventional transport/IP/MAC architecture. Furthermore, NC has been a prominent approach to improve the throughput of unicast traffic in wireless multi-hop networks through opportunistic NC. In this paper we seek a collaboration scheme for NC in multi-channel/interface wireless networks, i.e., the integration of NC, routing and channel assignment problem. First, we extend the NC for multiple unicast sessions to involve both COPE-type and a new proposed scheme named as Star-NC. Then, we propose an analytical framework that jointly optimizes the problem of routing, channel assignment and NC. Our theoretical formulation via a linear programming provides a method for finding source-destination routes and utilizing the best choices of different NC schemes to maximize the aggregate throughput. Through this LP, we propose a novel channel assignment algorithm that is aware of both coding opportunities and co-channel interference. Finally, we evaluate our model for various networks, traffic models, routing and coding strategies over coding-oblivious routing.