Experimental approach: energy-and-bandwidth spectrum sensing using GNU radio and USRP

Spectrum sensing is an essential process to implement dynamic spectrum access successfully and hence improve the spectrum utilization. This paper focuses on an experimental investigation of spectrum sensing using GNU radio and Universal Software Radio Peripheral (USRP) board. In the most related works, the method of energy detection is widely used for experiments on spectrum sensing, in which the energy amplitude of the received signal is the sole parameter to determine a channel's status. Specifically, if the amplitude of received energy exceeds a given threshold, the channel is detected busy, otherwise the channel is sensed idle. By observing experimental results, we find that the energy amplitudes of unwanted signals, which may be either interference or noise or both, are often at the same level as the desired signal. This will lead to a false alarm if the traditional energy detection is employed. At the same time, the bandwidth of unwanted signals are usually narrow and around 10KHz in a lab environment, which is far less than a regular signal's bandwidth, e.g. 40MHz in 802.11n and 20MHz in 802.11a. Inspired by these observations, we propose a new and simple approach termed Energy-and-Bandwidth Spectrum Sensing (EBSS), in which both the energy level and bandwidth of the received signal are taken into account. In addition, the sensing performance including false alarm and miss detection are analyzed. Extensive experiments are conducted and results verify that both false alarm and miss detection can be significantly reduced by the proposed approach.