Information Technology Reference
1.3 Model of ubiquitous node server
Ubiquitous computing, as mentioned, has caused a growth in the develop-
ment of algorithms for scheduling. All ubiquitous nodes assume that the data
required and sent are immediately available to the transmitter. This assump-
tion ignores the fact that the requested data should be taken from secondary
data repositories before being sent to other nodes. However, we consider that
these data are directly accessible to the node that has been requested.
The flow diagram in Figure 1 summarizes the structure of a generic
ubiquitous node which concerns the information exchange. The requests of
client nodes arrive at very small intervals of time (obvious consequence of
the fact that ubiquitous systems are intended to be used by the mass), and
since these requests cannot be instantaneously satisfied, they are queued. We
will call it as 'queue of the ubiquitous node', where all the pending requests
made by user nodes are placed. The information desired by the mass of
customers are typically contained in a disc or alternatively, as often happens,
in a cache memory in order to optimize the response speed of the system.
Once read from the (cache or mass) memory, data are instantaneously sent to
a transmitter, as explained earlier.
Initially, for simplicity, we assume that the communication channel has
an infinite speed of transmission, this means that the transmitter does not store
data to its input, and as soon as an object arrives, it is immediately transferred
to the user who had requested it. In the second stage, we will consider the
more realistic case where the channel introduces transmission delays, which is
equivalent to non-zero response times in transmission, and therefore, the creation
of queues in the input of the transmitter. We will see later the weight on the
overall performance of the system in the transmission queues. In any case, the
data sent through the communication channel reach the clients who have
requested them. (An object data is sent simultaneously to a multitude of users.)
Figure 1: Architecture of a server node.