Abstract: In this paper, we present a novel algorithm for detecting a bonded channel in a cognitive radio network where channel bonding is allowed for a higher data rate. The envelope detection algorithm is proposed to distinguish the center frequency of the received signal in order to determine whether the signal is transmitted by a primary user occupying a single channel or a secondary user occupying more than two channels when the channel is in use.
Introduction
The Medical Implant Communication System (MICS) band, spanning 402–405 MHz, serves as a dedicated frequency range for wireless medical implants. Standard regulations typically restrict these devices to a 300 kHz bandwidth. However, emerging healthcare applications, such as high-resolution real-time video from endoscopic capsules, demand significantly higher data rates than a single channel can provide.
To address this, "channel bonding" is employed, allowing secondary users to aggregate multiple idle channels into a single wider transmission pipe. In a Cognitive Radio (CR) framework, maintaining priority for "Primary Users" (standard medical implants) is essential. If a Primary User finds a channel occupied, it must determine if the occupant is a Secondary User that can be preempted. This research focuses on identifying whether a signal within a specific frequency slot belongs to a single-channel Primary User or is part of a larger bonded transmission by a Secondary User.
Proposed Methodology
The proposed Envelope Detection Algorithm leverages the principle of spectral symmetry. When a Primary User transmits, the signal is centered precisely within the individual channel slot. Conversely, a Secondary User centers its signal across the entire bonded block. Consequently, when observed within a single constituent channel slot, the Secondary User's signal peak appears offset from the center frequency.
The receiver performs downconversion of the target slot to baseband. An $N$-point Fast Fourier Transform (FFT) is then applied to the baseband signal to determine the location of the maximum spectral peak. If the peak is found at or near 0 Hz, the occupant is classified as a Primary User. A significant frequency offset indicates a Secondary User. To mitigate the effects of noise and signal fading, the algorithm employs a multi-window decision logic, integrating results over varying time intervals ($T=1, 2, 4$). A Secondary User is only confirmed if the frequency deviation exceeds a predefined threshold ($f_\lambda$) consistently across these multiple integration windows.
Results and Analysis
The algorithm was evaluated using MATLAB simulations of QPSK modulated signals with root-raised cosine pulse shaping. The Primary User occupied a 250 kHz bandwidth, while the Secondary User utilized a bonded block of four channels (1,150 kHz total bandwidth, including guard bands).
Performance was measured using the probability of false alarm ($P_f$) and missed detection ($P_m$). Results indicated that $P_f$ drops to near zero when the frequency threshold ($f_\lambda$) is set to 1.2 kHz or higher. Furthermore, the proposed algorithm demonstrated superior accuracy compared to a standard moving average baseline.
As shown in the comparison of the cost function ($0.5 P_m + 0.5 P_f$), the proposed method achieves a near-zero error rate at the optimal threshold, whereas the moving average method maintains a significantly higher error profile. The system also showed resilience at low Signal-to-Noise Ratios (SNR), suggesting it would be reliable in the electrically noisy environment of the human body.
Conclusion
This study presents an effective spectrum sensing technique for Wireless Body Area Networks (WBANs) that distinguishes between primary and bonded secondary users based on frequency offset detection. By exploiting the physical geometry of bonded protocols, the system ensures that critical medical devices can reclaim bandwidth from high-data-rate applications without interference. Future work will involve evaluating handover protocols in full network simulations and testing the algorithm on hardware platforms to account for real-world impairments like local oscillator drift.
Summary updated