Citation
Rokhani, Fakhrul Zaman and Kan, Wen Chih and Kieffer, John and Sobelman, Gerald E.
(2009)
Optimality of bus-invert coding.
IEEE Transactions on Circuits and Systems II: Express Briefs, 55 (11).
pp. 1134-1138.
ISSN 1549-7747; ESSN: 1558-3791
Abstract
Dynamic power dissipation on I/O buses is an important issue for high-speed communication between chips. One can use coding techniques to reduce the number of transitions, which will reduce the dynamic power. Bus-invert coding is one popular technique for interchip buses, where the dominant contribution is from the self-capacitance of the wires. This algorithm uses an invert line to signal whether the bus data are in its original or an inverted form. While the method appears to be a greedy algorithm, we show that it is, in fact, an optimal strategy. To do so, we first represent the bus and invert line using a trellis diagram. Then, we show that applying bus-invert coding to a sequence of words gives the same result as would be obtained by using the Viterbi algorithm, which is known to be optimal. We also show that partitioning an M-bit bus into P subbuses and using bus-invert coding on each subbus can be described as applying the Viterbi algorithm on a 2P-state trellis.
Download File
Additional Metadata
Actions (login required)
 |
View Item |
