Multi-dimensional spectrum map construction based on tensor completion scheme

Spectrum map, which is the key to solve the shortage of spectrum resource, can be characterized by many dimensions such as geographical location, frequency, time, and signal strength. Spectrum statement such as frequency occupancy, idle time, and signal coverage can be viewed from the spectrum map. However, in practice, due to the influence of compressed sensing or the noise from the acquisition process, the collected spectrum data is incomplete. In this paper, we propose a joint prediction model and the spectrum tensor completion method to recover the missing data. Based on the Universal Software Radio Peripheral (USRP), we conduct an experiment to collect the real-world spectrum tensor data for simulation. The simulation results show that the combined method has an improved completion performance.