Cooperative Spectrum Sensing (CSS)

CSS exploits geographically distributed sensing nodes to receive independent signal fading realizations. This spatial diversity combats multipath fading and shadowing effects that cripple single-node detection. When one node experiences a deep fade, others likely have strong signal reception, dramatically improving aggregate detection reliability. The diversity order scales with the number of cooperating nodes, though correlated shadowing among closely spaced sensors can reduce this gain.

A single cognitive radio may be shadowed by buildings or terrain, creating a hidden node that cannot detect an active primary user and causes harmful interference. CSS solves this by distributing sensors across different physical locations. Even if one node is shadowed, others with line-of-sight to the transmitter can detect the signal. This is the primary architectural motivation for cooperative sensing in cognitive radio networks.

A central fusion center collects local observations or decisions from all cooperating sensing nodes and applies a fusion rule to generate a global binary decision. The fusion center can operate on:

• Hard decisions: Binary 1/0 occupancy votes from each node
• Soft decisions: Raw test statistics or energy measurements
• Quantized soft decisions: Compressed statistics balancing performance and bandwidth The fusion rule (e.g., K-out-of-N, Likelihood Ratio Test) determines the final occupancy state.

CSS faces a fundamental design conflict: longer sensing durations improve probability of detection but reduce time available for secondary data transmission. This sensing-throughput tradeoff directly impacts secondary user capacity. Cooperative sensing can partially alleviate this by allowing shorter individual sensing times while maintaining detection accuracy through diversity combining. Optimal frame structures balance sensing duration, reporting overhead, and transmission time.

CSS networks face specific physical-layer attacks:

• Spectrum Sensing Data Falsification (SSDF): Malicious nodes report falsified measurements to corrupt the global decision
• Primary User Emulation (PUE): Attackers mimic primary user signals to force unnecessary spectrum evacuation
• Byzantine attacks: Coordinated adversaries strategically manipulate fusion outcomes Mitigation requires reputation management systems that assign trust scores based on historical reporting consistency.

The reporting channel between sensing nodes and the fusion center is often imperfect, subject to fading, noise, and bandwidth limitations. Imperfect reporting channels can corrupt even accurate local decisions before they reach the fusion center. Robust fusion rules must account for reporting errors through techniques like:

• Weighted combining based on channel quality
• Error-correcting codes on decision bits
• Censoring strategies where nodes with poor channels abstain from reporting

A central processing node that aggregates local observations from geographically distributed sensing nodes. The fusion center applies a fusion rule—such as the K-out-of-N rule or Likelihood Ratio Test (LRT)—to synthesize a global binary decision about spectrum occupancy. This architecture centralizes intelligence but introduces a single point of failure and requires robust reporting channels to mitigate errors from fading or interference.

Two fundamental strategies for transmitting sensing data to the fusion center. Hard Decision Fusion sends a single binary bit (occupied/vacant), minimizing bandwidth overhead but discarding signal confidence. Soft Decision Fusion transmits raw energy levels or quantized test statistics, preserving granularity for weighted gain combining and achieving superior detection sensitivity at the cost of higher reporting channel throughput.

A Byzantine attack where malicious nodes report falsified local sensing results to corrupt the global decision. Attackers may consistently report 'vacant' to cause interference or 'occupied' to trigger a denial of service. Mitigation relies on reputation management systems that assign dynamic trust scores based on the historical consistency of a node's reports with the consensus global outcome.

A fundamental design conflict in cognitive radio frame structure. Longer sensing durations improve the probability of detection and reduce the probability of false alarm, but directly consume the time slot available for secondary data transmission. Optimizing this tradeoff requires balancing the Neyman-Pearson Criterion for primary user protection against the secondary network's quality-of-service requirements.

Spatial diversity is the core benefit of cooperative sensing, exploiting geographically separated nodes to experience independent multipath fading. However, correlated shadowing occurs when nodes in close proximity encounter similar large-scale obstructions, degrading the expected diversity gain. Effective node placement and cluster-based architectures are critical to decorrelating observations and restoring detection reliability.

A fully decentralized alternative to fusion center architectures. Nodes exchange information iteratively only with one-hop neighbors and execute a consensus algorithm to converge on a common global decision. This eliminates the single point of failure and reduces infrastructure overhead, but introduces convergence latency and is sensitive to network topology dynamics.