edVirtus Reference

Who is Claude Berrou?

Claude Berrou (1951-): The Engineer Who Invented the Turbo-Code Principle

Claude Berrou is a French electrical engineer whose invention of turbo codes transformed channel coding. By combining parallel recursive systematic convolutional codes with interleaving and an iterative soft-decision decoder, he showed that practical systems could operate extraordinarily close to the limit predicted by information theory.

The breakthrough was publicly introduced in a 1993 paper with Alain Glavieux and Punya Thitimajshima. It did more than add another code to a catalogue: it established iterative exchange of probabilistic information as a general engineering strategy and reopened interest in message-passing methods throughout communications.

Electronics and Telecommunications in Brittany

Berrou was born in Penmarc'h, Brittany, on 23 September 1951. He trained in electronics and joined the Ecole Nationale Superieure des Telecommunications de Bretagne in 1978, working where integrated-circuit design and digital communications met.

His perspective was shaped by implementation. Coding gain has operational value only when the decoder fits available hardware, memory, latency, and power. Feedback and repeated refinement were familiar ideas in electronics, but they had not been exploited in channel decoders in the form Berrou envisioned.

A Different Route Towards Capacity

Claude Shannon's theorem established a channel capacity below which reliable communication is possible, yet constructive codes traditionally traded performance against decoding complexity. Algebraic block codes and convolutional codes provided reliable tools but remained visibly separated from the theoretical bound in demanding regimes.

Berrou pursued a compound system using simple constituent encoders rather than one overwhelmingly complex code. Random-like interleaving spread troublesome patterns, while iterative inference allowed each constituent decoder to use information discovered by the other.

The 1991 Patent and 1993 Disclosure

The fundamental French patent application filed in 1991 names Berrou as inventor of the turbo-code construction. At the 1993 International Conference on Communications, Berrou, Glavieux, and Thitimajshima jointly presented the first public paper and performance results.

Patent inventorship and research authorship record different contributions. Berrou originated the patented core; Glavieux was central to development and validation; Thitimajshima's doctoral work addressed the recursive systematic codes and parallel concatenation. The public breakthrough depended on turning conception into reproducible evidence.

Parallel Concatenation and Interleaving

A classic turbo encoder feeds the information sequence to two recursive systematic convolutional encoders. One sees the original order and the other sees a permuted order created by an interleaver. The transmitted symbols include the information and parity contributions from the two views.

The interleaver makes the combined code behave more like a long, irregular construction while retaining structured component decoders. Punctured codes can omit selected parity symbols to raise the rate, trading redundancy and coding gain against bandwidth.

Extrinsic Information and the Turbo Loop

Each soft-input soft-output decoder estimates bit probabilities. It sends the other decoder extrinsic information - the new evidence supplied by its parity constraints rather than a repetition of the received sample or prior message.

After interleaving or deinterleaving, the partner decoder uses that information as a refined prior. Several iterations can turn many weak individual estimates into a highly reliable codeword decision. Berrou chose the turbo name because the reuse of output information recalled feedback in a turbocharged engine.

An Extraordinary Claim Tested

The early simulation curves approached the Shannon-Hartley limit so closely that many experts suspected an error. Berrou and Glavieux themselves repeatedly checked the software, and independent researchers then reproduced and analysed the result.

Later work exposed qualifications as well as strengths: convergence depends on signal quality and code design; minimum-distance effects can create an error floor; long interleavers add delay; and each iteration consumes energy. Near-capacity performance is an engineering trade, not a free improvement.

Deployment and a New Research Direction

Turbo codes entered mobile, satellite, and deep-space systems. Improved forward error correction could be exchanged for lower power, greater coverage, higher rate, or a smaller link margin, making the method particularly valuable where energy and received signal strength were constrained.

The success also prompted researchers to revisit Robert Gallager's low-density parity-check codes, whose iterative decoding had been impractical in the 1960s. Factor graphs, message passing, iterative detection, and the broader turbo principle became central themes in modern receiver design.

Recognition and Legacy

Berrou and Glavieux received the IEEE Richard W. Hamming Medal, and Berrou received the Marconi Prize for turbo coding. He continued research and teaching at the institution now known as IMT Atlantique, extending iterative ideas in coding and information processing.

His lasting achievement is the turbo loop: let several constrained components exchange reliability rather than demand a perfect answer from one pass. The principle did not abolish the Shannon limit; it gave engineers a practical way to approach it and a new language for reasoning under uncertainty.