In mobile network, wireless links may be overloaded due to a high concentratioJl of mobile hosts in a given cell. As a result, the required QOS(Quality of Service) cannot be provided to all flows in that cell. So, there exists adaptive applications which can operate with various bandwidth levels. When a wireless link is overloaded, it can adapt by reducing the bandwidths of flows, is referred to as rate adaptation. In this paper, we analyze the performance of existing rate adaptation methods by simulations. The result of simulation experiments shows that average-fair adaptation scheme has very low network overhead but high bandwidth variance of flows, which causes the worse quality of service. We propose a rate adaptation method refined by combining fair adaptation and average-fair adaptation, and shows the refined method can provide lower bandwidth variance of flows by simulations.