Many recently proposed congestion control protocols rely on a precise knowledge of the actual link capacity to achieve high network utilization. However the link capacity varies in wireless networks and leads to performance degradation of current congestion control protocols. Such degradation is also characterized as underutilization and oscillatory behaviors. In this thesis, we propose the use of an Extended Kalman Filter (EKF) to make wireless congestion control protocols capable of perceiving the bandwidth variation. Our EKF-based Bandwidth Estimation (EBE) scheme can predict link capacity by monitoring the persistent queue size or congestion window size of a wireless link without the need for directly measuring the real-time bandwidth. We implement EBE in NS-2 and integrate it with XCP, VCP and CUBIC protocols. Through extensive simulation studies, we demonstrate significant performance improvements of both protocols in wireless networks as the result of using EBE.