What Are Multiuser Channels?

Multiuser channels are communication channel models that describe scenarios where a single shared medium carries signals from or to more than one user simultaneously. Unlike the classical point-to-point channel studied by Shannon, multiuser channels involve competing or cooperating transmitters and receivers whose rates cannot be characterized by a single capacity number. Instead, performance is described by a capacity region: the set of all jointly achievable rate tuples, one rate per user, under a given power and bandwidth constraint. The study of multiuser channels forms the core of multiuser information theory and provides the theoretical foundation for multiple-access wireless systems, broadcast networks, and relay communications.

The foundational models were identified in the 1970s and 1980s. Thomas Cover's survey of multiple user information theory systematized the principal channel models and their capacity results, establishing the multiple-access channel, the broadcast channel, the interference channel, and the relay channel as the four canonical cases around which the field is organized.

Multiple-Access and Broadcast Channels

The multiple-access channel (MAC) models the uplink scenario: many transmitters send independent messages to a single receiver. The capacity region of the MAC is fully characterized under Gaussian noise; it is a polyhedron whose corner points are achieved by successive interference cancellation, a technique in which the receiver decodes one user's signal, subtracts it from the received waveform, and then decodes the next. The broadcast channel (BC) is the dual: one transmitter sends separate messages to many receivers, which characterizes the downlink. Unlike the MAC, the capacity region of the general broadcast channel is not fully characterized, but for degraded Gaussian broadcast channels superposition coding achieves the boundary. The IEEE Transactions on Information Theory paper on the duality of Gaussian MAC and broadcast channels established a formal mathematical duality between the two, enabling results and algorithms derived for one to be translated to the other.

Interference Channels

The interference channel models two or more transmitter-receiver pairs sharing a common medium without coordination. Each receiver treats the other users' signals as interference unless a joint decoding scheme is arranged. The capacity region of the general interference channel remains an open problem in information theory, with exact results known only in special cases such as very strong interference, where the interfering signal is strong enough that each receiver decodes the other user's message first. In practice, the interference channel is the relevant model for cellular networks, unlicensed spectrum sharing, and cognitive radio systems, where multiple independent operators transmit in overlapping bands.

Capacity Bounds and Coding Strategies

Achievability results for multiuser channels rely on techniques such as random coding, superposition coding, and binning, while outer bounds are established through cut-set arguments and entropy inequalities. The IEEE survey of multi-way channels in information theory from 1961 to 1976 traced the development of these tools through the foundational period of the field, covering relay channels, two-way channels, and compound channels in addition to the canonical multiuser models. More recent work has extended these results to channels with channel state information at the transmitter, feedback, and secrecy constraints.

Applications

Multiuser channels have applications in a range of fields, including:

• Cellular wireless networks, where the MAC and BC models govern uplink and downlink resource allocation
• Satellite communications, where a single transponder carries signals from multiple ground stations
• Optical fiber networks, where wavelength-division multiplexing introduces multi-user interference at amplifiers
• Cognitive radio systems, where secondary users share spectrum with primary users under interference constraints
• Underwater acoustic networks, where shared channels must support multiple sensor nodes and relay nodes