ITR/SII Energy and Quality of Service Aware Ad-Hoc Networking
Dirk Grunwald, Timothy X Brown

We propose an integrated approach to energy efficient collaborative
communication in adhoc wireless networks.  Our objective is to reduce
communication power consumption while simultaneously increasing the
communication quality between the collaborators. Our model problem
assumes that people are interacting via adhoc networking integrated
into pocket computers for some common task. For the communication to be
successfully, all participants must be reachable via the network and
have sufficient energy to perform any necessary computation.  This
objective will be reached by developing protocols and algorithms in the
following research areas:

  o Link-layer controllers that adaptively learn low-power strategies
    for error correction codes, transmission power levels, and radio
    activity times for reduced overall power consumption, within the
    constraints of delay and error bounds.

  o Network-layer controllers that use inter-node attenuation,
    available node energy, and current traffic flows to optimize
    routing so as to maximize the availability of all nodes.

  o Network interfaces that allow the programmer to describe delivery
    constraints on messages such that communication schedules can reduce
    energy in a changing routing topology.

  o System level power modeling that allows the application layer to
    trade additional processor power to reduce networking power.

The project is targeted to the emerging spectrum of highly capable
pocket or wearable computers and the ad-hoc networks formed by those
computers. We will both deploy an experimental infrastructure and use
in-depth simulation models to evaluate our system. Our prototype
system will use a combination of laptop computers using 802.11
wireless networking and the advanced ``Itsy'' palm-top computer with
lower bandwidth (and significantly lower power) interfaces. Our
simulation models will use realistic whole-system power models for
advanced pocket computers coupled with a standard network simulator
that we have enhanced with more realistic RF propagation models.

This project is significant because it address a communication model
that will be common with future generation computation and
communication devices. It is realistic because it will use an
experimental network. It is general because we will be able to use the
simulation framework to model future compute and communication
networks. Finally, it is achievable because it leverages our on-going
work in optimizing channel assignment in wireless networks and energy
efficient computing.