This paper presents a system theoretic approach to, and performance analysis of improved congestion control methods i.e. the RED (Random Early Detection)-like AQM (Active Queue Management) algorithms. We revisit the principles of Fuzzy-RED and its linear counterparts and examine their advantages and possible shortcomings from a control theoretical point of view. The performance comparisons are based on extensive simulations. The major interest is on the requirement of a nonlinear control law.