Title :
A low power crosstalk-free bus encoding using genetic algorithm
Author :
Vinh, Truong Quang ; Kim, Young-Chul
Author_Institution :
Chonnam Nat. Univ., Gwangju
fDate :
March 31 2008-April 4 2008
Abstract :
In this paper, we present a new bus coding scheme based on genetic algorithm to eliminate crosstalk delay and to reduce power dissipation. This algorithm can be applied usefully for large width bus, in which it is difficult to find the optimal solution for bus coding. Our bus encoding scheme is capable to eliminate completely crosstalk problems in bus partitions as well as inter-partitions by using crosstalk-free codewords and active shielding wires. The energy-optimum code mapping for crosstalk-free codewords is chosen by the genetic algorithm based on the probability of transmission data. The experiment shows that the proposed coding technique can save the dynamic power up to 51% for 64-bit buses while totally avoiding the crosstalk delay.
Keywords :
crosstalk; encoding; genetic algorithms; system buses; bus partitions; crosstalk-free codewords; genetic algorithm; low power crosstalk-free bus encoding; power dissipation; Capacitance; Crosstalk; Delay; Encoding; Energy consumption; Genetic algorithms; Genetic engineering; Heuristic algorithms; Power dissipation; Power engineering computing;
Conference_Titel :
Computer Systems and Applications, 2008. AICCSA 2008. IEEE/ACS International Conference on
Conference_Location :
Doha
Print_ISBN :
978-1-4244-1967-8
Electronic_ISBN :
978-1-4244-1968-5
DOI :
10.1109/AICCSA.2008.4493518