Fusion Center

The fusion center collects local spectrum sensing reports from all cooperating secondary users via a dedicated control channel. Each cognitive radio transmits either a hard decision (a binary 0 or 1 indicating signal presence) or a soft decision (the raw energy level or likelihood ratio). The fusion center synchronizes these reports to a common time reference before applying the combining rule.

The fusion center applies a mathematical combining rule to synthesize individual observations into a global decision. Common rules include:

• OR Rule: Declares a primary user present if any single node detects it. Maximizes sensitivity but increases false alarms.
• AND Rule: Requires all nodes to agree. Minimizes false alarms but risks missing weak signals.
• K-out-of-N Rule: A flexible majority vote where at least K of N nodes must detect the signal.
• Likelihood Ratio Test: The optimal soft-combining method that weights each node's report by its local signal-to-noise ratio.

A single cognitive radio can be shadowed by buildings or terrain, causing it to miss a primary transmitter and falsely declare a spectrum hole. The fusion center overcomes this by exploiting spatial diversity—even if one node is in a deep fade, others likely have line-of-sight to the primary user. This is the primary architectural motivation for cooperative sensing and the fusion center's existence.

The control channel between cognitive radios and the fusion center is bandwidth-limited and subject to its own impairments. Censoring techniques reduce overhead by having nodes only report when they have high-confidence observations. Quantization compresses soft decisions into a few bits. In hostile environments, the reporting channel itself may be jammed, requiring error-correction coding or dedicated spectrum for control traffic.

Hard decision fusion transmits only a single bit per node, minimizing control channel bandwidth but discarding confidence information. Soft decision fusion transmits the full energy measurement or log-likelihood ratio, enabling near-optimal detection at the cost of higher reporting overhead. The trade-off is governed by the bandwidth-constrained detection theorem, which quantifies the performance gap between the two approaches.

The fusion center is a single point of failure and a high-value target for adversaries. Spectrum sensing data falsification (SSDF) attacks occur when compromised nodes send manipulated reports to corrupt the global decision. Byzantine defense mechanisms, such as reputation-based weighting and outlier detection algorithms, are implemented at the fusion center to identify and exclude malicious nodes before applying the combining rule.

The overarching architecture where multiple spatially distributed cognitive radios share their local detection results with a fusion center. This spatial diversity is the primary mechanism for overcoming the hidden node problem, where a single sensor is shadowed from a primary transmitter by a physical obstruction. By aggregating observations from geographically separated nodes, the network achieves a significantly higher probability of detection than any single radio could alone.

A fusion strategy where each cooperating node makes a local binary decision (signal present or absent) and transmits only that single bit to the fusion center. The center then applies a logical rule such as K-out-of-N, OR, or AND to reach a global verdict. This approach minimizes the bandwidth required for the control channel but discards valuable soft information about detection confidence.

A higher-fidelity fusion strategy where cooperating nodes transmit their raw energy measurements or log-likelihood ratios to the fusion center. The center applies statistical methods such as Equal Gain Combining (EGC) or Maximum Ratio Combining (MRC) to generate a global test statistic. This approach dramatically improves sensitivity in low signal-to-noise ratio environments at the cost of increased control channel overhead.

The primary vulnerability that cooperative sensing architectures are designed to solve. A cognitive radio is shadowed from a primary transmitter by terrain, buildings, or foliage, causing it to falsely declare a spectrum hole. If that radio begins transmitting, it causes harmful interference to a nearby primary receiver that it cannot detect. A fusion center mitigates this by correlating reports from nodes with unobstructed line-of-sight.

A Byzantine-style attack targeting the fusion center's decision logic. A malicious node transmits fabricated local sensing reports to corrupt the global decision. In a 'always-free' attack, the adversary reports the channel as vacant to trick the fusion center into enabling interference. In a 'always-busy' attack, it reports occupancy to deny legitimate access. Robust fusion rules with reputation-based weighting are required to isolate these adversaries.

The dedicated logical or physical channel over which cooperating cognitive radios transmit their local sensing reports to the fusion center and receive global decisions in return. This channel must be highly reliable and low-latency. In-band and out-of-band implementations exist, with the latter requiring a separate radio transceiver. The bandwidth and error rate of this channel directly constrain the fusion strategy that can be employed.