Achievable privacy-performance tradeoffs for spectrum sharing with a sensing infrastructure
Matthew Clark, Konstantinos Psounis · 2018
Motivated by growing demand for radio frequency spectrum, spectrum regulators are laying the groundwork for centralized, dynamic spectrum sharing systems that will enable efficient access to spectrum for new wireless technologies. With a database of spectrum user information, and an infrastructure of spectrum sensors, these sharing systems will leverage cognitive radio concepts to automatically identify suitable spectrum for users. Whether incumbent users should provide information directly to the database or rely on the sensor network for detection remains an open question with implications for the effectiveness of the sharing and the privacy of the users. Both methods present the potential to explore a privacy-performance tradeoff within the sharing system. In this work we assess this tradeoff with both sensing and interface obfuscation approaches in a spectrum sharing system. We identify key design parameters in a formal model for the sharing system architecture, and conduct a thorough simulation study of a real-world use case to quantify privacy and performance. While abstract models suggest sensing based solutions should compare favorably with obfuscation heuristics applied to the user interface, the performance of realistic sensor network designs suggests that achieving a favorable privacy-performance tradeoff with sensor networks may be significantly limited by practical considerations.