Information Technology Reference
In-Depth Information
(k)
(l)
F
IG
. 41. Continued.
the results revealed that the index allocation in the broadcast channels has an effect
in terms of both response time and energy consumption. It should be noted that an
increase in the number of channels increases both the amount of energy expended
in switching from one channel to another and conflicts in accessing data across the
parallel channels.
In order to address these issues, we developed and implemented different data
retrieval algorithms for parallel broadcast channels. In particular, we found that ex-
isting solutions for TSP problem were applicable to our problem. The results from
our study showed that when a small (less than about 45%) percentage of the objects
in a broadcast are requested, a TSP heuristic should be used to determine the access
order in which the broadcast objects should be retrieved. However, when a larger
percentage of the objects is requested, the Row Scan method should be used to avoid
the delay associated with switching between channels.
The retrieval scheme proposed reduces the number of passes over the broadcast
channels as the number of channels increases; however, this increases the number of
conflicts and consequently, an increase in the switching frequency among channels.
This results in an increase in the energy consumption. In an attempt to balance en-
ergy consumption and access latency, the scope of our research was extended by
developing a set of heuristics that generates an ordered access patterns, i.e., the
