Poisson channel

Poisson channel
Name	Poisson channel
Type	Optical communication / photon-counting channel
Introduced	1960s
Related	Photodetector, Shot noise, Optical fiber, Free-space optics

Poisson channel

The Poisson channel is a statistical model for communication systems in which counts of discrete events follow a Poisson process, developed in the context of optical and quantum-limited transmission. It connects foundations from Claude Shannon-style information theory with physical models studied by Roy J. Glauber, Hendrik B. G. Casimir, and experimental groups at institutions such as Bell Labs, Massachusetts Institute of Technology, and Bell Telephone Laboratories. The model underpins modern analyses by researchers affiliated with Bellcore, NASA, European Space Agency, and laboratories at Harvard University, Stanford University, and University of California, Berkeley.

Definition

The Poisson channel is defined by a mapping from an input signal to an output random counting process whose conditional distribution is Poisson; early formalizations appear alongside work by David Slepian, Thomas Cover, and Robert G. Gallager on channel coding. Inputs often represent optical intensities produced by sources like lasers from Bell Labs or transmitters used in experiments at California Institute of Technology and University of Cambridge, while outputs correspond to photon counts observed at detectors developed at MIT Lincoln Laboratory and Rutherford Appleton Laboratory. Practical considerations draw on device models by Herbert Kroemer and photonics groups at Philips Research and Hitachi.

Mathematical Model

Mathematically, the channel maps a nonnegative waveform or scalar x(t) or x to a counting process Y(t) or integer Y with conditional law Poisson(lambda(x)), a structure employed in analyses by Leonard Kleinrock and in queueing contexts studied by Agnes Pepis and John R. Pierce. For memoryless variants the output distribution satisfies P[Y = k | X = x] = e^{-λ(x)} λ(x)^k / k!, with λ often affine in x to model background light characterized in experiments at NASA JPL and European Southern Observatory. Continuous-time formulations use inhomogeneous Poisson processes connected to work by Henry C. Tuckwell and stochastic calculus techniques developed by Kiyosi Itô and Paul Lévy. Extensions incorporate dark counts, dead time, and afterpulsing modeled in device literature from Hamamatsu Photonics and sensor characterizations at Lawrence Berkeley National Laboratory.

Capacity and Information-Theoretic Properties

Capacity results for Poisson channels build on variational techniques advanced by Claude Shannon, Robert M. Fano, and later by Imre Csiszár and János Körner; specialized bounds and achievability proofs rely on methods used by Elwyn Berlekamp and Andrew Viterbi. Both unconstrained and power-constrained capacities have been characterized in works involving researchers at Princeton University, Columbia University, and University of Illinois Urbana-Champaign. The role of constraints such as peak constraints, average power, and cost functions connects to optimization studies by George Dantzig and convex analysis by Jerome Keeler; converse proofs use techniques akin to those in the coding theorems of David Blackwell and Peter Elias. Quantum-optical generalizations relate to capacity problems studied by groups at Institute for Advanced Study and in theoretical physics literature referencing John Preskill and Alexander Holevo.

Estimation and Detection in Poisson Channels

Estimation theory for Poisson channels leverages classical signal-processing frameworks by Harry Nyquist and detection theory developed by Norbert Wiener and W. E. Zurek; maximum-likelihood and Bayesian estimators have been applied in work from Bellcore and AT&T Laboratories. Hypothesis testing with Poisson observations uses likelihood-ratio tests that mirror Neyman–Pearson insights by Jerzy Neyman and Egon Pearson and is implemented in sensor networks researched at Sandia National Laboratories and Los Alamos National Laboratory. Estimators for intensity functions use nonparametric techniques from Bradley Efron and regularization methods popularized by David Donoho and are used in imaging research at Argonne National Laboratory and Lawrence Livermore National Laboratory.

Coding and Modulation Techniques

Practical coding for Poisson channels adapts pulse-position modulation, on–off keying, and intensity modulation developed in optical communications research at Bell Labs, Toshiba Research, and Nokia Bell Labs. Error-correcting codes such as Reed–Solomon codes and low-density parity-check codes from Irving S. Reed and Robert Gallager have been combined with modulation schemes tested by teams at Ecole Polytechnique Fédérale de Lausanne and Telecom Paris. Rate-adaptive schemes and feedback strategies draw on results by Thomas Cover and Yihong Wu; implementations have been prototyped in free-space optical links by groups at Caltech and MIT Lincoln Laboratory.

Applications and Examples

Applications span deep-space optical communication pursued by NASA JPL and quantum-limited optical receivers in quantum key distribution demonstrations by groups at University of Geneva and Los Alamos National Laboratory. Photon-counting LIDAR systems used in surveys by US Geological Survey and atmospheric sensing by National Oceanic and Atmospheric Administration exploit Poisson models. Biomedical imaging modalities like single-molecule fluorescence studied at Max Planck Institute for Biophysical Chemistry and Cold Spring Harbor Laboratory rely on Poisson statistics; astronomical photometry at European Southern Observatory and Space Telescope Science Institute models photon arrivals as Poisson processes. Sensor networks in ecology and remote sensing deployed by Smithsonian Institution and Woods Hole Oceanographic Institution also apply Poisson-based models.

Category:Information theory