builds an energy efficient cellular network infrastructure. Currently, the wireless cellular infrastructure is experiencing surging energy consumptions. GreenBS aims to improve energy efficiency in base station (BS) subsystem, the most critical subsystem that is the dominant contributing factor to overall energy consumption. Our work shows that
Base stations (BSes) in the 3G cellular network are not energy proportional with respect to their carried traffic load and 3G traffic exhibits high fluctuations both in time and over space, thus incur energy waste. We proposed a profile-based approach to green cellular infrastructure. First, they profile BS traffic and approximate network-wide energy proportionality using non-load-adaptive BSes. The instrument is to leverage temporal-spatial traffic diversity and node deployment heterogeneity, and power off under-utilized BSes under light traffic. We showed that this simple scheme can yields up to a large portion of energy savings (e.g. 53% in a dense large city and 23% in a sparse, mid-sized city). More information about GreenBS can be found in Mobicom'11 paper.
aims to provide a scalable wireless interconnect for many-core systems-on-chip. It explores a new direction of designing wireless shortcuts on many-core system on chip. This project uses wireless networking and data center technology to renovate the network infrastructure and associated solutions in a futuristic systems on chip of thousands of cores.
I did the early feasibility study and literature survey on this topic. More information can be found in Mobicom'09 paper.
This work was done when I was an associate researcher at MSRA. More information about Point&Connect can be found in this page, or Mobisys'09 paper.
seeks a high-accuracy ranging technique, without leveraging any
pre-planned infrastructure or special hardware/system modification. It should
be a software-based solution and uses only the most basic set of commodity
hardware so that it is readily applicable to many low-cost sensor platforms and
COTS mobile devices. The key idea is to use the pair-wise the elapsed time of
arrival (ETOA) of sound signals between the two devices to infer the distance.
We have implemented BeepBeep prototype on Windows 5.0+ mobile phones and demonstrated that it can achieve around one or two centimeters accuracy
within a range of more than ten meters. Our BeepBeep demo won Best Demo Award at Sensys'07.
This work was done when I was an assistant researcher at MSRA. More information about BeepBeep can be found in this page or Sensys'07 paper.
addresses the channel allocation problem in the context of open spectrum access (cognitive radio). We formulate it to a variant of the graph coloring problem and show the global optimization problem is NP-hard. We develop the centralized and distributed scheme to provide an approximation solution through vertex labeling. Our experimental results show that our allocation algorithms can dramatically reduce interference and improve throughput (as much as 800%). Further simulations show that our distributed algorithms can work effectively as the centralized ones, with much less computational complexity.
This work was done when I was an intern at MSRA with Dr. Haitao Zheng. It is also part of project Nautilus. More information can be found in ICC'05 and ACM Monet'06 paper.
Last update: 02/04/2012