A scalable array for Cellular Genetic Algorithms: TSP as case study

Abstract

Cellular Genetic Algorithms (cGAs) exhibit a natural parallelism that makes them interesting candidates for hardware implementation, as several processing elements can operate simultaneously on subpopulations shared among them. This paper presents a scalable architecture for a cGA, suitable for FPGA implementation. A regular array of custom designed processing elements (PEs) works on a population of solutions that is spread into dual-port memory blocks locally shared by adjacent PEs. A travelling salesman problem with 150 cities was used to verify the implementation of the proposed cGA on a Virtex-6 FPGA, using a population of 128 solutions with different levels of parallelism (1, 4, 16 and 64 PEs). Results have shown that an increase of the number of PEs does not degrade the quality of the convergence of the iterative process, and that the throughput increases almost linearly with the number of PEs. Comparing with a software implementation running in a PC, the cGA with 64 PEs has shown a 45x speedup.

A framework for hardware cellular genetic algorithms: An application to spectrum allocation in cognitive radio

“…Catapult HLS version 2010a (University Version) and Precision RTL 2010a were used to synthesize the PE and cGA controller, and Xilinx ISE and EDK 13.4 to implement the complete embedded system. The global RNG has been implemented with a cellular automata ring network as in [7]. The target device is a Xilinx Virtex-6 FPGA (XC6VLX240T-1) on a ML605 board.…”

“…This shows an interesting potential for acceleration with coarse-grain parallel processing architectures, as several solutions can be evolved simultaneously. In previous work we validated the concept of the proposed array architecture with the traveling salesman problem [7]. In this paper we propose a flexible and scalable hardware framework for supporting this category of GAs and present implementation and execution results for a relevant optimization problem.…”

Automated framework for FPGA-based parallel genetic algorithms

“…In addition to a considerable number of conventional GA systems mentioned in [4], [5], [6], [20], FPGA-based master-slave GAs and dGAs have been demonstrated [7], [8], [9], [10], [11]. FPGA-based cGAs are also proposed in [12], [13], [14]. In the top 6 rows of Table II we summarise the features of these existing pGA systems.…”

“…There have been previous attempts to adapt pGAs to FPGAs for acceleration or low power consumption [3], [4], [5], [6], [7], [8], [9], [10], [11], [12], [13], [14]. However, designing an FPGA-based pGA is not as easy as implementing multiple hardware blocks supporting a set of GA instances.…”

An Empirical Analysis on Dimensionality in Cellular Genetic Algorithms

“…al. proposed a toroidal array of Processor Elements (PEs) which deal with small sub-populations each; however, toroidal connection is lost among individuals and therefore the canonical algorithmic structure of a cellular GA is significantly modified [8,19]. Cellular GAs are massively parallel GAs which ideally can evolve one generation in few clock cycles.…”