Abstract
This paper introduces SEEMA (Spectrum and Energy Efficient Multiple Access), a protocol designed for efficient event detection in Wireless Sensor Networks (WSNs). SEEMA addresses the dual challenges of spectrum scarcity and energy constraints by employing a censoring strategy where only sensors with highly informative data transmit. By utilizing an analog Multiple Access Channel (MAC), SEEMA enables simultaneous transmissions, ensuring that bandwidth requirements remain independent of the network size. Theoretical analysis and simulations demonstrate that SEEMA achieves exponential error decay and superior energy efficiency compared to traditional Time Division Multiple Access (TDMA) schemes.
Introduction
In typical WSN detection tasks, a Fusion Center (FC) must distinguish between hypotheses (e.g., detecting a gas leak versus normal conditions) based on data from $N$ sensors. Traditional communication approaches, such as TDMA, allocate individual time slots to each sensor, which leads to two major issues: linear scaling of bandwidth requirements and significant energy waste from transmitting uninformative data. SEEMA is proposed as a solution that achieves high efficiency while maintaining simple hardware requirements for individual sensors.
Methodology
The SEEMA protocol is built upon three fundamental pillars:
- Censoring: Sensors only transmit if their local observation falls within a specific "informative" region ($\Gamma_n$). If a sensor's data is ambiguous, it remains silent, significantly saving energy.
- Analog Multiple Access Channel (MAC): Instead of sequential transmission, all active sensors transmit a common waveform simultaneously. The FC receives a superposition (sum) of these signals, making the bandwidth requirement independent of the number of sensors.
- Simplified Sensor Design: Sensors do not need global statistical knowledge; they simply follow a threshold rule provided by the FC, which handles the complex mathematical processing.
Performance is analyzed using Large Deviation (LD) theory and the Gartner-Ellis Theorem to prove that the probability of detection error decays exponentially as the network grows, even under non-ideal conditions like fading or spatial correlation.
Results and Analysis
Numerical simulations in MATLAB across various scenarios—including i.i.d. Gaussian signals, non-i.i.d. observations, and Markovian correlated data—validate the protocol's effectiveness. A critical aspect of SEEMA's performance is the relationship between the censoring threshold and the resulting error exponent.
As illustrated in the figure above, the error exponent varies based on the normalized expected number of transmitting sensors. The performance peaks at an optimal threshold (e.g., $X_L = \theta/2$), demonstrating that transmitting only the most informative data maximizes the detection rate. When the threshold is too high ($X_L \to \infty$), too few sensors transmit; when it is too low ($X_L \to -\infty$), the inclusion of uninformative data degrades the exponent.
Key insights from the results include:
- Bandwidth Invariance: Unlike Type-Based Multiple Access (TBMA), SEEMA’s bandwidth remains fixed regardless of data dimensionality.
- Noise Resilience: By transmitting simultaneously, the FC only deals with a single instance of channel noise, whereas TDMA accumulates noise from every individual sensor's time slot.
- Scaling: Error probability decays exponentially with $N$ even when individual sensor transmission energy scales down as $1/N$.
Conclusion
SEEMA provides a scalable and energy-efficient framework for detection in WSNs. By leveraging the superposition property of the wireless medium and intelligent censoring, it decouples bandwidth needs from network size while maintaining high detection accuracy. Future work will address practical implementation challenges, such as achieving precise phase synchronization among low-cost sensors and optimizing censoring regions for diverse statistical distributions.
Summary updated