A modular analog NLMS structure for system identification

Adaptive filters have been mainly developed in a digital way which may involve, in some applications, a large number of computations for filtering and adaptation. This fact may restrict the use of adaptive filters in some applications like channel equalization in cellular systems in which high performance, fast convergence rates and small size ADF are required. This paper proposes an analog adaptive filter algorithm in which the filter adaptation is carried out by using a normalized analog LMS algorithm. Proposed structure has, potentially, a very small size and faster convergence rates than its digital counterparts. This fact makes it suitable for several applications such as equalization and echo cancellation in cellular telephones in which high performance with low power consumption systems are required