Decision Making for Two-Phase Network-Coded Cooperative Information Repair in Wireless Networks

In our previous work [6], we proposed a cluster based two-phase PPIR protocol to repair the lost packets among end users during the 3G Base Station (BS) broadcast. The PPIR protocol achieves better performance compared with others since cluster heads have the full knowledge of each cluster and collect the data packets efficiently by exchanging control information with cluster members during the first phase. However, such a scheme incurs a large communication overhead for control information exchange when the cluster size scales up, which leads to large repair delay. In this paper, we design a new scheduling scheme to reduce this overhead by reducing the number of control information exchange. In this manner, cluster heads do not have the full knowledge of their own clusters and need make the decision when to stop waiting for more data packets and enter the second phase. To solve such decision making problem, we propose two approaches, named as the optimization based approach and the optimal stopping approach to reduce the total repair latency. In the first approach, we formulate the problem based on a defined cost metric and find the stopping time according to the minimum cost. In the second approach, every time the cluster head receiving a data packet, it makes a decision based on the knowledge of already received data packets. Simulation results show that the two proposed approaches achieve smaller repair latency compared with the PPIR protocol, and the first approach is suitable for the case of low packet loss rate (during the BS broadcast), while the second approach is suitable for the case of high packet loss rate.