Adaptive and occupancy-based channel selection for unreliable cognitive radio networks

In this paper, we propose an adaptive and occupancy-based channel selection for unreliable cognitive radio networks.

💡 Research Summary

**
The paper addresses the fundamental challenge of channel uncertainty in Cognitive Radio Networks (CRNs) by proposing an Adaptive Occupancy‑Based Channel Selection (AOCS) algorithm. Traditional CRN channel‑selection schemes typically treat channel availability as a binary condition (available or not) and rely on periodic spectrum sensing to pick a channel. In realistic environments, however, primary user (PU) activity is highly dynamic, sensing errors are inevitable, and hardware imperfections or interference further obscure the true occupancy state. Ignoring these uncertainties leads to excessive retransmissions, increased latency, and reduced throughput.

Problem Modeling
The authors first model each channel’s occupancy probability (OP) using a Bayesian estimator that fuses prior observations with current sensing outcomes. The estimator explicitly incorporates a sensing error probability (ε) and a two‑state Markov model for PU activity (active/inactive). This probabilistic representation captures both the stochastic nature of PU behavior and the imperfections of the sensing process.

Weighted Occupancy (WO) Metric
To avoid relying solely on the instantaneous OP, the paper introduces a Weighted Occupancy (WO) metric:

 WO_i(t) = α·OP_i(t) + (1‑α)·R_i(t)

where R_i(t) is the empirical reliability of channel i, measured as the historical success rate of transmissions on that channel, and α∈