Network coding is seen as a promising technique to improve network throughput. In this paper, study two important problems in localized network coding in wireless networks, which only requires each node to know about and coordinate with one-hop neighbors. In particular, establish a condition that is both necessary and sufficient for useful coding to be possible. Content verification is an important and practical issue when network coding is employed. When random linear network coding is used, it is infeasible for the source of the content to sign all the data, and hence, the traditional “hash-and-sign” methods are no longer applicable. However, this technique is difficult to be applied to network coding designed for high calculation and message overhead. It explores this issue further by carefully analyzing different types of overhead, and proposes methods to help reducing both the computational and communications cost, and provide provable security at the same time. It show this condition is much weaker than expected, and therefore permits a change of coding schemes to suit different network conditions and application preferences. Based on the understanding establish, able to design a robust coding technique called loop coding that can improve network throughput and TCP throughput simultaneously..