Unimodal biometric systems remain constrained by noisy acquisition, intra-class variability, non-universality, and susceptibility to spoofing, limitations that multibiometric systems address by combining evidence from two or more traits. Among the available integration strategies, feature-level fusion is widely regarded as the most information-rich, since it operates on the raw extracted feature sets before any matching decision is made, but it is also the most difficult to implement because of the curse of dimensionality, structurally incompatible feature spaces, and the absence of a known optimal combination rule. Metaheuristic search has emerged as a practical route through this difficulty, and the Bacteria Foraging Optimization Algorithm (BFOA) has attracted growing attention on account of its parallel search structure, insensitivity to initialization, and demonstrated effectiveness in feature selection tasks, despite well-documented weaknesses in convergence speed and premature stagnation. This paper presents a systematic literature review of feature-level fusion techniques and BFOA-based optimization in multibiometric authentication, covering studies published between 2013 and 2026 and drawn from IEEE Xplore, ScienceDirect, SpringerLink, MDPI, PubMed, and Google Scholar. Thirty-eight primary studies satisfied the inclusion criteria after a PRISMA-guided screening process. The review synthesizes the evolution of BFOA variants, chemotaxis-step and swarming modifications, and hybridizations with Lévy flight, particle swarm optimization, and chaotic search, alongside parallel developments in immune-based and evolutionary feature-selection methods applied to face, fingerprint, and iris fusion. The synthesis shows that although recognition accuracies exceeding 97% have been reported by several fusion frameworks, very few studies apply BFOA specifically at the feature level of a trimodal face-fingerprint-iris system, and fewer still address the algorithm’s slow convergence and premature-stagnation weaknesses in that context. Building on these gaps, the review outlines a research direction for an Enhanced BFOA (EBFOA) that adapts the chemotactic step size across iterations and applies a non-uniform elimination-dispersal probability to preserve population diversity, intended to improve both recognition accuracy and computational efficiency in feature-level multibiometric fusion. The paper closes by identifying open questions for future primary research, including benchmark standardization, cross-dataset generalization, and real-time deployment feasibility.

Keywords: Bacteria Foraging Optimization Algorithm, feature-level fusion, multibiometric systems, multimodal biometrics, metaheuristic optimization, systematic literature review.