The use of adaptive antenna systems enables the improvement of the overall wireless network capacity, especially in Mobile Ad hoc Networks (MANET) environments in which the quality of the communication is closely related to the Signal-to-Noise-Ratio (SNR) measured in the channel. Unfortunately, most of used adaptive algorithms such as Least Mean Square (LMS) and Constant Modulus Algorithm (CMA), because are strongly related to the step size value and to the error metric estimation, usually present a speed convergence problem. In this paper, a new Variable Error Metric (VMA) adaptive Algorithm is proposed in order to limit the level of dependency between the algorithms and step size also reducing the slowing effect introduced by Mean Square Error (MSE) criterion used for example by LMS algorithm.