What Is Channel Capacity?

Channel capacity is the theoretical maximum rate at which information can be transmitted over a communication channel with an arbitrarily small probability of error. The concept was established by Claude Shannon in his 1948 paper "A Mathematical Theory of Communication," which founded the discipline of information theory. Shannon showed that every channel, regardless of the underlying physical medium, is characterized by a capacity C measured in bits per second, and that reliable communication is possible at any rate below C but impossible at any rate above it. This result, known as the Shannon channel coding theorem, separates the limits of reliable communication from the specific coding schemes used to approach those limits.

Channel capacity is determined by two physical properties: the bandwidth available for transmission and the signal-to-noise ratio (SNR) at the receiver. The Shannon-Hartley formula C = B log₂(1 + S/N), where B is bandwidth and S/N is the signal-to-noise power ratio, gives the capacity of a band-limited channel corrupted by additive white Gaussian noise (AWGN). The formula shows that capacity grows linearly with bandwidth and logarithmically with SNR, a relationship with profound consequences for how engineers design high-throughput communication systems.

The Shannon Limit and Coding Theory

The channel coding theorem proved that codes exist that approach capacity but did not specify how to construct them. For decades after Shannon's 1948 paper, the gap between the best practical codes and the Shannon limit remained substantial. The development of turbo codes by Berrou, Glavieux, and Thitimajshima in 1993, and the subsequent analysis of low-density parity-check (LDPC) codes, demonstrated that practical codes could operate within a fraction of a decibel of the Shannon limit. The full text of Shannon's original 1948 paper, hosted at Harvard, remains a foundational reference and establishes the vocabulary that all subsequent information-theoretic analysis inherits.

The noise model matters: the AWGN result applies to channels dominated by thermal noise, but capacity formulas for fading wireless channels, binary symmetric channels, and channels with memory differ in their derivation and interpretation. Ergodic capacity averages over fading states when the channel is stationary and ergodic; outage capacity characterizes scenarios where the fading is slow and the transmitter lacks full channel knowledge.

Capacity of Multi-Antenna and Multiuser Channels

The discovery that multiple-input multiple-output (MIMO) antenna systems can multiply channel capacity by the number of independent spatial streams was a turning point in wireless communications. In a rich scattering environment, a system with min(M, N) transmit-receive antenna pairs achieves a capacity that scales approximately as min(M, N) times the single-antenna capacity at the same total power. The MIT News explanation of the Shannon limit and practical approaches to achieving it places MIMO capacity gains in historical context alongside the development of turbo and LDPC codes. Broadcast and multiple access channels, analyzed through the theory of network information theory, have distinct capacity regions that describe the achievable rate tradeoffs among multiple users sharing the same medium.

Quantum Channel Capacity

The extension of channel capacity theory to quantum communication channels introduces new phenomena with no classical analog. Quantum channels can transmit classical information, quantum information (qubits), and entanglement, and separate capacity formulas apply to each resource type. The quantum channel capacity for transmitting quantum information is bounded by the coherent information of the channel, a quantity that can be negative when the channel is very noisy. Some quantum channels exhibit superadditivity, meaning that the capacity per channel use increases when the channel is used multiple times in parallel with entangled inputs. These properties are studied within quantum information theory, which draws on the mathematical framework of density operators and completely positive maps. The information theory course notes from UC Berkeley's Redwood Center cover both classical and quantum channel capacity in a unified treatment. Within quantum communications, channel capacity for qubit transmission is bounded by results derived in the quantum Shannon theory literature developed since the mid-1990s.

Applications

Channel capacity has applications in a range of communications engineering and information-theoretic contexts, including:

• Design of forward error-correcting codes for digital communications and storage
• Link budget analysis and modulation selection in wireless network planning
• Spectral efficiency optimization in LTE, 5G NR, and Wi-Fi standards
• Capacity analysis of quantum key distribution and quantum communication links
• Compression limit analysis in data storage and source coding