Innovating the Capability of Mobile System Using Radio Technology.

摘要

Recent emergence of smart devices has opened the unprecedented era of mobile computing, which is now the vital part of our daily lives. Yet, a few hurdles remain that limit the functionality of mobile systems. In this dissertation we have identified three main challenges. The first problem is poor mobile throughput in random access networks. This is mainly caused by networking overheads, such as backoffs or packet collisions, when multiple mobile users attempt to access the network. Current singletransmission based network protocols fail to effectively deal with such networking overhead. Another problem is under-utilization of frequency spectrum. While the frequency spectrum is scarce resource, it is not being well utilized due to absence of fast and energy-efficient broadband surveying technique. Finally, indoor localization is yet inefficient. Existing techniques call for an extensive process of site profiling, which is labor-intensive and time-consuming. Therefore the wide deployment of the indoor location-based services is being hindered. We strive to innovate the capability of the mobile system in the three aforementioned aspects. With our expertise in radio techniques, we design, implement and test the real system and prove the effectiveness of our solutions. The solutions are titled Contrabass, QuickSense, and ACMI. Contrabass is a concurrent transmission protocol designed to improve the network throughput. Physical and MAC layer protocols of Contrabass are deliberately designed to eliminate the explicit negotiation between the stations. Therefore the networking overhead is minimized and the mobile communication becomes highly efficient. QuickSense is a broadband spectrum sensing technique based on off-the-shelf analog hardware. QuickSense implements the binary search based spectrum survey and helps efficient utilization of the spectrum. We show that QuickSense has the time complexity as low as that of the compressed sensing. Finally, we present ACMI, an FM-based indoor localization that does not require proactive site profiling. ACMI automates the construction of RSS fingerprint database based on the estimation of indoor RSS distribution. ACMI further performs multi-level online signal matching to cope with the estimation errors. As a result, ACMI achieves good indoor localization accuracy without the site profiling efforts. We outline the contribution of this dissertation as follows. We first define three hard challenges in the mobile computing area. Then we present novel but practical solutions that tackle the challenges. Each solution is designed and implemented toward the complete mobile system. Finally, the systems have been extensively tested and proven successful.