Communication in the Presence of Noise
Authors: Claude E. Shannon
Year: 1949
Source: Proceedings of the IRE, 37(1), 10–21
Contributions
- Nyquist-Shannon sampling theorem: a bandlimited signal can be perfectly reconstructed from discrete samples taken at a rate ≥ 2× the maximum frequency (the Nyquist rate)
- Formal proof of channel capacity via the Shannon-Hartley theorem
- Unified continuous and discrete communication theory within a single mathematical framework
- Established the mathematical basis for all modern digital communications (audio, video, networking)
Methodology
Mathematical analysis using Fourier theory and probability. Extends Shannon’s 1948 paper to continuous-time channels with additive white Gaussian noise (AWGN). The sampling theorem is proved rigorously by showing that any bandlimited function can be expanded as a series of sinc interpolations at the Nyquist rate. Channel capacity is derived by treating the noise as a Gaussian random process and applying the entropy results from information theory.
Results
- Sampling theorem proved: a signal bandlimited to B Hz is fully determined by 2B samples per second
- Capacity formula for the AWGN channel: C = B log₂(1 + S/N), where B is bandwidth, S is signal power, and N is noise power
- Demonstrated that increasing bandwidth or signal-to-noise ratio increases capacity, but with diminishing returns under log growth
Limitations
- Assumes stationary Gaussian noise — real channels often have non-stationary or non-Gaussian interference
- Relies on infinite-bandwidth idealization for some derivations
- No practical encoding scheme is provided; the theorem is an existence result about theoretical capacity
- Perfect reconstruction assumes exact Nyquist-rate sampling; in practice, oversampling and anti-aliasing filters are required