Cooperative Sequential Spectrum Sensing Based on Level-Triggered Sampling

Abstract

We propose a new framework for cooperative spectrum sensing in cognitive radio networks, that is based on a novel class of nonuniform samplers, called the event-triggered samplers, and sequential detection. In the proposed scheme, each secondary user (SU) computes its local sensing decision statistic based on its own channel output; and whenever such decision statistic crosses certain predefined threshold values, the secondary user will send one (or several) bit of information to the fusion center (FC). The FC asynchronously receives the bits from different SUs and updates the global sensing decision statistic to perform a sequential probability ratio test (SPRT), to reach a sensing decision. We provide an asymptotic analysis for the above scheme, and under different conditions, we compare it against the cooperative sensing scheme that is based on traditional uniform sampling and sequential detection. Simulation results show that the proposed scheme, using even 1 bit, can outperform its uniform sampling counterpart that uses infinite number of bits under changing target error probabilities, SNR values, and number of SUs.Index Terms-Asymptotic optimality, cognitive radio, decentralized detection, event-triggered sampling, randomized quantization, sequential probability ratio test (SPRT).

Sequential Distributed Detection in Energy-Constrained Wireless Sensor Networks

“…In this paper, we accordingly propose to encode overshoot into the time-delay between sampling time and transmission time. We show that the result of [9] on the order-2 asymptotic optimality holds for the proposed scheme.…”

“…This approach achieves the order-1 asymptotic optimality, but not order-2, which is a stronger type of asymptotic optimality. In [9], at each time t k n some additional bits are used to quantize the overshoot q k n . This approach can achieve the order-2 asymptotic optimality if the number of quantization bits increases at a rate of log | log γ| where γ → 0 at least as fast as the error probabilities α and β, i.e., γ = O(α), γ = O(β).…”

“…To tackle the overshoot problem, in [8] and [10], the log-likelihood ratio of the bit received by the FC is computed via simulations, which includes an average value for the overshoot. Alternatively in [9] the overshoot in each sample is reported to the FC at the sampling time in a quantized form, resulting in a multi-bit scheme. The former approach achieves the order-1 asymptotic optimality, whereas the latter achieves the order-2 asymptotic optimality if the number of quantization bits increases at a reasonably low rate with the decreasing target error probabilities.…”

“…In particular, in the level-triggered sampling [8], [9], the local LLR L k t is sampled at a sequence of random times {t k n } n that are dynamically determined by L k t itself. Specifically, the nth sample is taken at time t k n when the change in the signal since the last sampling time t k n−1 exceeds a predetermined threshold ∆, i.e.,…”

“…Whereas in [9], at time t k n , in addition to b k n , sensor k also transmits some additional bits quantizing the overshoot q k n , and then the FC computesq k n and accordingly, from (5), an approximate valueλ k n of λ k n using the received bits. Since we have…”