Efficient and low-cost software defined radio on commodity hardware

As general purpose processors become faster, many signal processing functions that used to require special purpose hardware, are now feasible to perform in software. In wireless communications, this means we move towards the “ideal” software-defined radio (SDR), where we have an antenna and analog/digital converters, after which the signal is processed in software. Defining most of the functionality in software makes software defined radios very flexible, in that they can be reconfigured quickly to perform any task for which they have sufficient computational capacity. This thesis is composed of a number of studies into high-performance SDR, and very low-cost SDR. The first series of studies explore how well modern programmable general purpose processors are able to run typical signal processing tasks in the physical layer of a modern communications standard: DVB-T2. In the second series of studies, a radio spectrum monitoring system is built using very low-cost components, and some software. The first series of studies was initiated by an analysis of the DVB-T2 standard for digital television broadcasting. The most computationally complex signal processing functions of a DVB-T2 receiver were implemented on a graphics processing unit (GPU), and various central processing units (CPUs). The objective of these studies was to show that a real-time capable receiver for a modern standard such as DVB-T2 is feasible to implement completely in software on modern off-the-shelf computers. The second series of studies focuses on distributed radio frequency spectrum monitoring using very low-cost sensors. The sensors are based on cheap off-the-shelf hardware, which was never intended for spectrum monitoring. The nodes were instead adapted to the tasks of spectrum sensing and geolocation of transmitters by running specialized software.