PRECISE Seminar: Wireless Coexistence in Open Radio Spectrum: Curses and Blessings

The human world is replete with wireless devices. By their broadcast nature, wireless transceivers often cause significant interference to each other if they use the same frequency. This becomes a growing issue as the open 2.4GHz frequency band is being populated by numerous wireless devices including Wi-Fi access points and handhelds, ZigBee sensors, Bluetooth headsets, and cordless phones. In this talk, I will present our recent work on the coexistence of Wi-Fi and ZigBee. Our study of real-life Wi-Fi networks reveals that abundant /white space/ exists in Wi-Fi traffic. We have developed a novel approach that enables ZigBee wireless links to achieve /assured/ performance by exploiting such white space. Although the coexistence of different wireless technologies has been traditionally considered a curse, in our latest work, we have exploited it as blessings. We developed a new sensor network time synchronization approach called WizSyc where ZigBee sensors can detect and synchronize to the periodic beacons broadcasted by WiFi access points, without resorting to multi-hop message passing. WizSyc has been implemented in TinyOS and extensively evaluated on a real wireless testbed.

In the second part of this talk, I will discuss two on-going projects on collaborative sensing in cyber-physical systems. In the first project, we propose a new approach for profiling harmful diffusion processes like oil spills and chemicals leaks, using a smart aquatic mobile sensor platform called robotic fish. In our approach, robotic fish collaboratively profile the characteristics of a diffusion process (source location, discharged substance amount, evolution over time) and coordinate their movement to progressively improve the profiling accuracy. In the second project, we are developing a sensor network system for real-time, in-situ, and long-term volcano detection and tomography computation. By employing novel in-network collaborative signal processing algorithms, our system can meet stringent requirements on volcanic sensing quality at low power consumption.