A Dell desktop functions as the sink node. We should note
that we are only concerned with the wireless segment from the source node to
the relay node, thus the connection from the relay node to the sink is wired.
We tested three types of RIs on the source sensor node: D-Link DWL-650,
Linksys WPC11, and Lucent/Orinoco Gold. A fourth RI was used to capture
tra?±c in promiscuous mode using tcpdump. We also used the iwconfig utility
to record the transmission rate of the RI in one second intervals. For each
experiment, the sensor source node sends data over the wireless link through
the relay node to the sink. The test is repeated ten times for each RI.
We imposed a tra?±c load of 2.4Mbps. The load was light enough to not
stress the node, but heavy enough to cause the RI to be the bottleneck in the
sensor so that it is the primary factor influencing the behavior of the tra?±c.
We used the sock [6] program to establish a user datagram protocol (UDP)
connection carrying constant bit rate (CBR) tra?±c generating a 1470-byte
packet every five milliseconds.
In a real environment, rate switching occurs due to the changes in channel
conditions caused by noise. This noise may be caused by the network contention,
interference from neighboring networks operating on the same channel,
mobility of a wireless sensor, non-802.11 devices operating in the same
frequency range, collisions with link layer acknowledgements, etc.
Pages:
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680