Bayesian Risk-Based Sequential Detection for Fast Color-Sequence Identification in Optical Camera Communication

Zuni-Mango R.; Coaguila-Chavez E.; Tamo A.; Hilario-Tacuri A. · IEEE Access (2026) · DOI: 10.1109/access.2026.3730381

Resumen

Optical Camera Communication (OCC) systems commonly employ predefined color sequences for transmitter identification, localization, and beacon-based communication. In these applications,the receiver must identify the transmitted sequence as quickly as possible without compromising detectionreliability. Conventional approaches typically rely on either a fixed number of observations or heuristicconfidence thresholds, neither of which explicitly addresses the trade-off between detection reliabilityand observation latency. We present a Bayesian risk-based sequential detector for fast color-sequenceidentification in OCC systems. Sequence identification is formulated as a Bayesian sequential decisionproblem, and a one-step look-ahead approximation is used to determine whether acquiring an additionalobservation is expected to improve the decision. The proposed detector jointly accounts for the risk ofincorrect sequence identification and the cost of delaying the decision, allowing the observation time toadapt to the accumulated statistical evidence. The main novelty lies in replacing fixed observation lengthsand heuristic confidence thresholds with a risk-based stopping criterion that explicitly balances identificationerrors against observation latency. Its performance is evaluated through Monte Carlo simulations underdifferent observation noise conditions and compared with fixed-repetition maximum a posteriori (MAP)and posterior-threshold detectors. The results show that the proposed method maintains reliable detectionwhile reducing the average observation time over a wide range of operating conditions, making it suitablefor latency-sensitive OCC applications requiring rapid and reliable color-sequence identification.

Instituciones: Universidad Nacional de San Agustín de Arequipa