Bit-Energy Comparison of Discrete and Continuous Signal Representations At the Circuit Level

We employ the formalism of information theory, and investigate the performance, as measured in bitenergy, of continuous/discrete-value and continuous/discrete-time signal representations as realized in electrical circuits (communication channels). We derive the signal-to-noise ratio, average power dissipation, Shannon channel capacity and bit-energy. Bit-energy, in Joules per bit, is a reasonable goodness criterion to compare information ow in physical channels where encoding, transmission and decoding of messages must be carried out under strict power supply/dissipation constraints. Without loss of generality, we assume that signals are voltages which are limited by the power supply rails in the circuits and we use simple but realistic models of noise and power dissipation.

[1]  R. W. Sanders Communication Efficiency Comparison of Several Communication Systems , 1960, Proceedings of the IRE.

[2]  Andreas G. Andreou,et al.  On physical models of neural computation and their analog VLSI implementation , 1994, Proceedings Workshop on Physics and Computation. PhysComp '94.

[3]  B. Logan Information in the zero crossings of bandpass signals , 1977, The Bell System Technical Journal.

[4]  B.J. Hosticka,et al.  Performance comparison of analog and digital circuits , 1985, Proceedings of the IEEE.

[5]  Richard E. Blahut,et al.  Principles and practice of information theory , 1987 .