Optimal Routing, Link Scheduling
and Power Control in Multi-hop Wireless Networks
We are generally interested
in the problem of optimizing the timing and duration
of sleep states on mobile devices, with the objective
of minimizing power with respect to a QoS constraint.
In our current model, there are two power consumption
modes, sleep and active. There is associated rate
of power consumption at the mobile terminal in
each state, as well as a fixed energy consumed
in transitioning between the two states. The QoS
parameter we are focused on is
average delay.
To help gain a better understanding
of the general problem, we considered a simple
model where there is a single transmitter and
receiver. The receiver is the mobile node whose
mode we wish to
control. The transmitter can give commands to
the receiver regarding its sleep state, and forwards
incoming streaming data to the receiver appropriately.
We formulated this as a Markov decision process,
and solved it numerically using dynamic programming.
The solutions from the numerical calculations
strongly suggest that the optimal policy (that
which minimizes average power consumption subject
to an average delay constraint) is such that the
transmitter should only command the receiver to
sleep when there is no data queued at the transmitter.
The system thus behaves as a single server queue
with vacations. We were able to derive closed
form expressions for the optimal sleep duration,
as well as the associated minimal rate of power
consumption. Future work will be focused on more
elaborate models involving multiple users.
