How does spectrum sensing work in cognitive radio?

Spectrum sensing detects whether a primary user is currently transmitting on a given frequency. Three main techniques are used. Energy detection measures the received power in a frequency band and compares it to a threshold: if the power exceeds the threshold, a primary user is assumed present. This is the simplest technique but suffers from poor performance at low SNR (below -10 dB) and cannot distinguish primary signals from noise or interference. The sensing time for reliable detection at SNR = -10 dB with probability of detection Pd = 0.9 and false alarm Pfa = 0.1 is approximately 10 to 100 ms depending on bandwidth. Cyclostationary feature detection exploits the periodicity in modulated signals (caused by carrier frequency, symbol rate, or cyclic prefix) that noise does not possess. It can detect primary signals at SNR as low as -20 dB but requires knowledge of the primary signal's cyclic frequencies and has higher computational complexity. Matched filter detection correlates the received signal with a known primary signal template, achieving optimal detection performance (SNR as low as -25 dB) but requiring complete knowledge of the primary signal format. In practice, cooperative sensing combines results from multiple cognitive radios to overcome shadowing and multipath effects that can hide a primary signal from a single sensor.

How does cognitive radio avoid interfering with primary users?

Multiple mechanisms work together. Spectrum sensing provides the first line of defense: the cognitive radio continuously monitors for primary user signals and must vacate the channel within a specified time (2 seconds in IEEE 802.22, 60 seconds in CBRS) upon detection. Geolocation database access provides a second mechanism: the cognitive radio queries a database of primary user locations and operating parameters, and the database returns a list of available channels and maximum permitted transmit power at the cognitive radio's location. This eliminates the hidden node problem where a cognitive radio cannot detect a primary receiver. Transmit power control limits the cognitive radio's EIRP to keep interference at primary receivers below a threshold (typically -100 to -110 dBm per channel at the primary receiver). Beamforming and directional transmission further reduce interference by concentrating energy toward the intended cognitive receiver and away from primary users. Dynamic frequency selection (DFS) automatically moves the cognitive radio to a different channel when a primary signal is detected, similar to how Wi-Fi in the 5 GHz band avoids radar signals. The combination of these techniques achieves primary user protection with interference probability below 0.1% in well-designed systems.

Wireless Protocols

Cognitive Radio

Q: What is the CBRS framework?

The Citizens Broadband Radio Service (CBRS) is a three-tier spectrum sharing framework established by the FCC in the 3550 to 3700 MHz band (3.5 GHz) in the United States. Tier 1 is Incumbent Access (IA), protecting existing users including US Navy shipborne radar systems and fixed satellite stations. Tier 2 is Priority Access License (PAL), sold via auction in 10 MHz channels per county with interference protection from Tier 3 but required to vacate for Tier 1. Tier 3 is General Authorized Access (GAA), available to anyone with certified equipment on a best-effort basis with no interference protection. The system is managed by a Spectrum Access System (SAS) that maintains a database of incumbent locations and protection zones, receives Environmental Sensing Capability (ESC) sensor reports of incumbent activity, and dynamically assigns channels to PAL and GAA users. CBRS devices must contact the SAS every 300 seconds to maintain their channel assignment and must vacate within 60 seconds if the SAS revokes access due to incumbent detection. Maximum EIRP is 47 dBm for Category B (outdoor) devices.

/kog-nih-tiv ray-dee-oh/

Cognitive radio senses the electromagnetic environment and dynamically accesses unused spectrum (white spaces) without interfering with licensed primary users. Cycle: sense → decide → share → vacate. Sensing techniques: energy detection (-10 dB SNR), cyclostationary (-20 dB), matched filter (-25 dB). Commercial: IEEE 802.22 (TV white spaces), CBRS (3.5 GHz, three-tier SAS-managed). Protection: <0.1% interference probability.

Category: Wireless Protocols

CBRS: 3550 to 3700 MHz

Vacate time: 2 to 60 s

Understanding Cognitive Radio

Spectrum scarcity is largely an illusion. Measurements consistently show that most licensed spectrum is unused most of the time in most locations: occupancy studies in major cities find that only 15 to 25% of allocated spectrum below 6 GHz is actively transmitting at any given time. The remaining 75 to 85% sits idle, reserved by license but not utilized. Cognitive radio exploits this inefficiency by allowing unlicensed devices to use the idle spectrum opportunistically, creating a massive increase in effective spectrum capacity without requiring new frequency allocations.

Joseph Mitola III introduced the concept in 1999, and Simon Haykin formalized the engineering framework in 2005. The core idea extends software-defined radio (which can tune to any frequency and use any modulation) with awareness and intelligence: the radio not only can operate on different frequencies, it knows which frequencies are available and decides which to use based on learned environmental knowledge. This requires continuous spectrum monitoring, real-time decision-making, and rapid frequency agility to vacate channels when primary users return.

Spectrum Sensing Performance

Energy Detection Threshold:
λ = σ_n² (1 + 1/√N) × Q^-1(P_fa)

Sensing Time (energy detection):
N ≈ 2 [Q^-1(P_fa) - Q^-1(P_d)(1+SNR)]² / SNR²

Cooperative Sensing Gain:
P_d,coop = 1 - ∏(1 - P_d,i) (OR rule)

Where σ_n² = noise variance, N = samples, P_fa = false alarm probability, P_d = detection probability. At SNR = -10 dB, P_d = 0.9, P_fa = 0.1: N ≈ 2,500 samples. 5 cooperating radios at P_d = 0.5 each: P_d,coop = 0.97.

Dynamic Spectrum Access Frameworks

Framework	Band	Access Model	Vacate Time	Max EIRP
IEEE 802.22 WRAN	54 to 862 MHz	Sensing + database	2 s	36 dBm
CBRS (US)	3550 to 3700 MHz	SAS database	60 s	47 dBm (Cat B)
TV White Space	470 to 790 MHz	Geolocation DB	10 s	36 dBm
LSA (EU)	2300 to 2400 MHz	Licensed shared	Minutes	Varies
AFC (Wi-Fi 6E)	5925 to 7125 MHz	AFC database	N/A	36 dBm (SP)

Common Questions

Frequently Asked Questions

How does spectrum sensing work?

Energy detection: compare received power to threshold (works to -10 dB SNR, 10 to 100 ms sensing time). Cyclostationary: exploits modulation periodicity (to -20 dB, higher complexity). Matched filter: correlates with known signal template (to -25 dB, requires signal knowledge). Cooperative sensing from multiple radios overcomes shadowing: 5 radios at P_d=0.5 each achieve P_d,coop=0.97.

What is the CBRS framework?

Three-tier sharing at 3.5 GHz. Tier 1 (Incumbent): Navy radar, protected. Tier 2 (PAL): auctioned 10 MHz channels per county. Tier 3 (GAA): open access, no protection. SAS manages assignments; devices check every 300 s and vacate in 60 s. Max EIRP: 47 dBm outdoor. Enables private LTE/5G networks.

How is interference avoided?

Layered protection: spectrum sensing + geolocation database + power control + beamforming + DFS. Database eliminates hidden node problem. Power control keeps interference <-100 dBm at primary receivers. Cognitive radio must vacate in 2 s (802.22) or 60 s (CBRS). Combined: <0.1% interference probability.

Related Terms

← Cognitive Radar Cognitive Radio Defense →