|
Exploiting Channel
Variation in Scheduling Delay-sensitive Data Flows
Issues: Wireless channel conditions are time-varying.
Such a variation can be either slow or very fast due to the different level
of Doppler frequency. In this project, we take the channel variation into
consideration and address the issue of scheduling delay-constrained data
packets.
Our Approach and Main Findings: We first notice problems
for existing two approaches when they are used to schedule delay-sensitive
data flows over time-varying channels.
- The greedy algorithm
works well in fast time-varying channels. It is proved to achieve ½ worst-case
performance ratio. However, in slow time-varying situations, it may
drop data packets with urgent deadlines.
- EDF
(earliest-deadline first) achieves optimality when the channel is slow
time-varying. But if the variation is fast, EDF becomes inefficient because
it does not take advantage of the transient peak
transmission rate.
To guarantee a high performance of the scheduling algorithm in both slow
and fast time-varying situations. We propose an adaptive algorithm which is
insensitive to the variation speed of wireless channels. In the proposed
algorithm, the scheduler employs channel prediction technique to determine
a range of time period in which the channel conditions won’t change
much, then the schedule uses an EDF-alike strategy to schedule packets
within this range. Beyond the range, the proposed algorithm switches to a
greedy-alike strategy. The proposed algorithm is proved to achieve a
worst-case performance ratio higher the greedy algorithm. The most important advantage of this
algorithm is that its performance adapts to the channel variation
automatically. Our simulations have confirmed the results.
Project Members:
- Students: Xiaoqiao Meng
- Faculty: Songwu Lu
Publications:
- Xiaoqiao
Meng, Songwu Lu.
Exploiting Channel Variation in Scheduling Delay-sensitive Data Flows.
Under submission, July, 2003.
|
