Information Technology Reference
In-Depth Information
bacKground
wake up before that time. Many researches on
mobile broadcast focus on designing broadcast
index (Hu, 1999; Imielinski, 1994; Lee, 2003; Lo,
2000; Shivakumar, 1996; Tan, 2000).
mobile broadcast
According to the mechanism used for schedul-
ing data objects on the broadcast channel, data
broadcast can be classified into three categories:
push-based, pull-based, and hybrid. In push-based
broadcast (Acharya, 1995), the clients do not in-
teract with the server: the server broadcasts data
objects based on its knowledge of the mobile peers'
overall data access requirements, and the mobile
peers get their required data objects when they
appear on the broadcast channel. In pull-based
broadcast (Aksoy, 1998), which is also called on-
demand broadcast, the mobile peers submit their
requests through an uplink channel to the server,
then the server broadcasts data objects based on
the received requests (pull-based broadcast is dif-
ferent from point-to-point access in that the object
requested by a query is delivered on a broadcast
channel such that all the pending queries for that
data object are answered with one transmission).
Hybrid broadcast (Acharya, 1997) combines push-
based broadcast and pull-based broadcast (or even
point-to-point access) to complement each other:
popular data objects are pushed, and infrequently
accessed data objects are retrieved on demand (by
pulling or point-to-point access).
Data objects are typically scheduled for broad-
casting according to the overall access interests of
the mobile peers in the system. The basic idea is
to broadcast hot data objects more frequently than
the others. Many algorithms have been proposed
for scheduling the broadcast in different environ-
ments (Acharya, 1998; Guo, 2001; Hameed, 1997;
Liu, 2003; Zheng, 2005).
To facilitate mobile peers saving energy, index
of data objects is introduced into the broadcast.
Basically, a broadcast index tells the mobile peers
what data objects will be available when. The
broadcast index itself is also broadcast intermit-
tently on the broadcast channel. Once a mobile
peer knows when its interested data object will
be broadcast, it can operate in sleep mode and
System model
In this chapter, we assume a broadcast environ-
ment where:
•
The mobile broadcast is push-based. A server
broadcasts data objects repeatedly to many
mobile peers through a broadcast channel
and there is no uplink for the mobile peers
to send requests to the server.
•
The broadcast is non-uniform. Data objects
are broadcast with different frequencies.
Popular data objects are broadcast more
frequently than the others.
•
Broadcast index is available so that mobile
peers know the broadcast frequencies of
data objects.
•
All mobile peers are in the broadcast server's
transmission range. They get their required
data objects by tuning into the broadcast
channel.
•
All data objects are of the same size and
they are not updated. This assumption lets
us ignore the factor of object size and the
problem of cache invalidation so that we
focus on the cooperative strategies for cache
management.
•
The mobile peers are equipped with short
range communication devices. A mobile
peer can communicate with the mobile peers
within its communication range.
•
There is no super peer in the mobile peers.
Peers have the same levels of computational
ability and battery power.
Sample applications
Many interesting applications follow the system
model that we described before. In such applica-
Search WWH ::
Custom Search