ABSTRACT
This chapter presents the HGA, a genetic algorithm written in VHDL and intended for a hardware implementation. Due to pipelining, parallelization, and no function call overhead, a hardware GA yields a significant speedup over a software GA, which is especially useful when the GA is used for realtime applications, e.g., disk scheduling and image registration. Since a general-purpose GA requires that the fitness function be easily changed, the hardware implementation must exploit the reprogrammability of certain types of field-programmable gate arrays (FPGAs), which are programmed via a bit pattern stored in a static RAM and are thus easily reconfigured.
After presenting some background on VHDL, this chapter takes the reader through the HGA’s code. We then describe some applications of the HGA that are feasible given the state-of-the-art in FPGA technology and summarize some possible extensions of the design. Finally, we review some other work in hardware-based GAs.
