A Cooperative Routing Algorithm to Increase QoS in Wireless E-Healthcare Systems - E-Healthcare Systems and Wireless Communications - page 384

nodes can seriously affect the dynamic conges-
tion mainly by inducing flash crowd phenomena.
Herein we set up an evaluation environment
in TrueTime (Ohlin, 2007). The simulation
parameters include: path loss exponent 3, sen-
sor transmission power -3dbm, receiver signal
threshold -48dbm, sink buffer length 200pkt,
sensor buffer length 30 pkt, retry limit 3, data rate
250Kbps and collision avoidance mechanism of
the ZigBee protocol, maximum reporting rate of
sink 50 pkt/sec. In the following section, we de-
fine some of the performance metrics considered
in simulation experiments. Reliability is defined
as the percentage of total sent packets that are
received at the sink.
Moreover, the memory limitation of the sink
node implies packet dropping when the incoming
traffic heavily exceeds the maximum sink capacity.
We note that the number of collisions and MAC
Layer Errors can affect packet losses in wireless
networks. To take into account the above factors
influencing the packet dropping, we have evalu-
ated the average total packet (in percentage) loss
metric, including packets lost due to both buffer
overflow and collision/Mac layer errors.
Several health care critical applications require
that the observed event (or control action) is reli-
ably detected (actuated) within a certain time limit.

Hence, where it is of interest, we have considered
the average latency (delay) of data packets to
study the trade-off related to latency. Additionally,
energy efficiency is evaluated. Indeed, in WBAN,
energy efficiency is also crucial due to the limited
energy resources of the sensors. Finally, the per-
formance of the proposed algorithm is evaluated
in a mobile node scenario.
There now follows a set of simulation scenarios
used to validate the proposed QBAR algorithm.

Static Scenarios

We first consider the nominal scenarios in Figure
2 where sources S ia , S ib and S ic send packets to the
sink. We show the effectiveness of the proposed
algorithm in balancing the load by using alterna-
tive paths. In contrast, in the standard AODV
algorithm the path to the destination is static even
if congestion occurs with the resulting buffer
overflow and network performance degradation.
As shown in Figures 3-4, the reliability and
the responsiveness obtained using QBAR is 35%
higher than with AODV. In addition, the average
packet delay is lower. QBAR switches the traffic
between the intermediate nodes avoiding buffer
overflow. In contrast, AODV mainly uses S i3 and
S i5 with a resulting buffer overflow. Moreover,

Figure 2. Nominal Simulation scenarios