Wireless Networking Research
Self-Managing
Chaotic Wireless Networks
Until recently, most dense deployments of wireless networks were in campus-like environments, where experts carefully plan cell layout. The rapid deployment of cheap 802.11 hardware and other personal wireless technology (2.4Ghz cordless phones, bluetooth devices, etc.), however, is quickly changing the wireless landscape. The resulting dense deployment of wireless networking equipment in areas such as neighborhoods, shopping malls, and apartment buildings, differ from campus-like deployments in two important ways. First, while campus deployments are carefully planned to optimize coverage and minimize cell overlap, many recent deployments result from many independent people or organizations each setting up one or a small number of APs. This type of spontaneous, unplanned deployment results in highly variable (including very high) densities of wireless nodes and APs. Moreover, 802.11 nodes share the spectrum with other networking technologies (e.g., Bluetooth, UWB) and other uses (e.g., cordless phones). Second, configuring and managing wireless networks is difficult, and most users do not have the skills to deal with even simple tasks such as choosing parameters such as SSID and channel, let alone more complex questions such as power control abd number and placement of APs. We have coined the term chaotic networks to refer to this type of deployments. This project is developing protocols and techniques for making chaotic networks self-configuring and self-managing. For example, we are developing an platform that can collect a real time signal map of the environment to help us better understand how interference affects wireless network performance.
Publications:
Team members:
Retired team members:
Research
aimed at evaluating and improving wireless network protocols is hindered by the
inability to perform repeatable and realistic experiments. Evaluating wireless
protocols is challenging because, while
the physical layer can often be ignored in wired networks, the physical layer
in wireless networks fundamentally affects operation at all layers of the protocol
stack. Traditional techniques for
evaluating networking protocols include testbeds and simulation. Testbeds offer a high degree of realism, but
wireless testbed experiments suffer from a lack of repeatability and
control. Simulation does provide full
control and repeatability, but achieving realism is difficult because the
simulator must model all components in the system and their interactions.
We are developing a wireless emulator that
enables realistic and repeatable wireless experimentation by accurately
emulating wireless signal propagation in physical space. The emulator takes in
the RF signals generated by wireless network cards, subjects them to the same
effects that occur in real physical space (e.g. attenuation, multi-path fading,
…), and feeds the combined signals back into the wireless cards. The heart of the system is an FPGA that
executes realistic, real-time signal propagation models. Unlike previous approaches, this emulator
uses a real MAC layer and a realistic physical layer without adopting an
uncontrollable or locale-specific architecture.
Research areas include the architecture of the emulator, the software
environment for specifying and managing experiments, emulator-based evaluation
methodologies, and the characterization of emulation-based evaluation of
wireless networks relative to simulation and testbeds.
More information can be found on the project’s web page.
Publications:
Team members:
Retired team members:
Video
Streaming over Wireless Networks
The motivation for this project is that entertainment is likely to be a significant application for residential wireless networks. However, while there has been a significant amount of research on video streaming over wireless, high-quality video over wireless remains an elusive goal. There are a number of reasons for this. First, the aggressive compression and the nature of video coding (use of inter-frame dependencies) make video very sensitive to packet loss. Another problem is that the real-time nature of video means that there is a deadline for frames: frames that arrive late are equivalent to a lost frame from the perspective of the receiver. Even worse, the transmission of late frames results in wasted network resources. We are currently pursuing a technique called Content-aware Adaptive Retry (CAR) in which retransmission in the 802.11 MAC layer is optimized based on an understanding of the structure of the MPEG video stream. Specifically, for each Group Of Pictures (GOP), frames are sent in order of importance (I, B, and P) and frames that cannot be transmitted before the end of the GOP time interval are discarded. This includes ``cutting short'' the retransmissions of frames. Initial results for CAR are very promising.
Publications:
Team members:
Today's devices use radios where most of the functionally is implemented in hardware. The idea behind software radios is to instead use software that processes the digitized signals. The advantage of this approach is that it provides greater flexibility. For example, different wireless standards can be implemented on the same hardware by installing different software packages; this is potentially a significant advantage for devices that need to connect to different types of wireless networks. This flexibility also makes software radios an attractive platform for communications and networking researchers since it is possible to change both the physical and MAC layer functions of the device; these functions are typically implemented in hardware or proprietary software in current commercial devices.
This project uses the Universal Software Radio Peripheral (USRP) platform for research in flexible and dynamic wireless nodes. The USRP is a fairly basic software radio component that includes A/D and D/A converters and an FPGA. It can be used with a variety of frontends, which are attached to the motherboard as daughter cards. The GNU Radio framework is used for software development. We are using the USRP for projects ranging from new MAC protocols to the development of “spectrum sensors” that allow nodes to learn about their environment so they can adapt.
Team members:
Retired team members:
An Internet-style Approach to Wireless
Networks (1996-2003)
While wired links are isolated from sources of interference, including other links, wireless links share the spectrum, making them susceptible to interference from other links and external uncontrolled radio sources. As a result, wireless links are error-prone and have variable properties (e.g. bandwidth, delay, loss rate). This affects all layers of the protocol stack. For example, TCP can have very poor performance as a result of non-congestion packet losses and quality of service guarantees cannot always be met. This project uses an Internet-style approach to address these problems: it applies localized solutions when possible (e.g., losses) and for wireless features that cannot be hidden inside the wireless layer it provides interfaces so higher layers in the protocol stack and observe and control the behavior (e.g., QoS).
Publications:
Retired team members:
Last updated, August 21, 2005, Peter Steenkiste