Practical approaches for spectrum sensing using software defined radios
Abstract
Current communication systems face growing demand for high data rate and low latency. One way to satisfy such demand is to use full spectrum resources because licensed frequency spectrum resources are limited and underutilized. For this matter, spectrum sensing techniques enable radio devices to search and use vacant spectrum regardless of the possession of a license. The major problem with existing spectrum sensing techniques is that most are required to have heavy computational algorithms and extensive prior-knowledge. It causes either high computation or inefficiency to implement. Even if a few techniques achieve low computation or high efficiency, they are impractical to implement due to low detection rates or high false alarm rates. Our work aims to provide and implement practical sensing solutions with software-defined radio (SDR) which guarantee high flexibility to control and build major functions of radio systems. Our work also considers different characteristics of a licensed band and an unlicensed band. In the unlicensed band, especially the 2.4GHz ISM band, a number of non-WiFi devices cause Wi-Fi traffic outages and often complete disconnection. Detection and identification of these types of interference are necessary to avoid interference problems. Our research simply uses spectrum power information provided by WiFi access points and performs classification to identify non-WiFi interference. This method achieves over 90% success rates and below 3% false classification rates. In addition to this classification work, we develop a simple interference detection scheme using various detection algorithms that include the Kolmogorov-Smirnov test. In the licensed band, such as UHF TV band, most existing spectrum sensing works are associated with simple energy detection. To enable unlicensed users to access vacant licensed bands without any possible conflict, smart spectrum sensing algorithms should be applied more than just a simple detection of the licensed user. In our work integrating with the DARPA spectrum challenge, the detection scheme first analyzes the noise floor of the SDR device to set the energy detection threshold. Based on this analysis, we implement a quantized spectrum sensing algorithm to detect white space or transmit-able space in the spectrum. Instead of simply detecting energy presence, it also analyzes the bandwidth of the signal and determines the radio transmission.
Degree
Ph.D.
Advisors
Krogmeier, Purdue University.
Subject Area
Electrical engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.