This paper presents an efficient method for the hand off mechanism in cellular networks using optimization algorithms. The proposed approach integrates a fuzzy logic approach with simulated annealing algorithm to automate the tuning process. The fuzzy controller carries out inference operation at high-speed, whereas the tuning procedure works at a much lower rate. For the implementation described in this paper, a two-input-one-output fuzzy controller is considered. Both the inputs and the output have 8- bit resolution, and up to seven membership functions for each input or output can be defined over the universe of discourse. The fuzzy controller has two levels of pipeline which allows overlapping of the arithmetic as well as inference operations. The SA tuning mechanism adjusts the triangular or singleton membership functions to minimize a cost function. The complete self-tuned fuzzy inference engine is implemented in a Xilinx SPARTAN3 XC3S200 series FPGA device. This paper describes various aspects of the implementation of the self-tuned hand off system.