Performance Evaluation of Reservation MAC Protocols
239
both extended basic polling and extended active polling achieve the best results in highly
loaded network.
In low and middle load areas, both combinations of basic and active polling with pig-
gybacking ensure shorter signaling delays than protocols without piggybacking. The best
results in the low load area are achieved by both extended protocols, which additionally
apply the extended random access to data channels to be used for signaling including
the dynamic backoff mechanism and piggybacking. However, near the network saturation
point, (300 stations) the extended protocols do not improve the performance compared
with basic and active polling with piggybacking.
In networks with rare transmission requests, the signaling delay decreases in the high
network load area if active polling or piggybacking are applied (Fig. 6.32). The combi-
nation of active polling and piggybacking achieves again the best results, as is the case
in the network with frequent requests. However, the influence of active polling is more
significant if longer user packets are transmitted. The reason for this is a correspondingly
longer absence of dedicated request slots for the stations sending or waiting for access
rights, which is caused by the longer transmission times necessary for larger packets. This
decreases the general round-trip times of the polling messages and reduces the signaling
delay as well. Reduction of the signaling delay in the low network load area is also
achieved by application of both extended protocols.
Generally, it can be concluded that the combination of active polling and piggybacking
improves the network utilization, ensuring the theoretical maximum value in both net-
works with frequent and rare transmission requests. This protocol combination reduces
the signaling delay significantly in high network load area. However, the signaling delay
in the middle load area is still longer than in the case of protocols with random access in
networks with rare requests (e.g. extended ALOHA, Sec. 6.3.2).
10
100
1000
50
100
150
200
250
300
350
400
450
500
Signaling delay (ms)
Number of stations
Basic polling
Polling
+ piggy
Active
+ piggy
Extended act. poll.
Extended polling
Active
polling
Polling
+ piggy
Active
polling
Active
polling
Figure 6.32
Mean signaling delay polling-based reservation MAC protocols rare requests
(average packet size: 1500 bytes)
Summary :
Performance Evaluation of Reservation MAC Protocols 239 both extended basic polling and extended active polling achieve the best results in highly loaded network. Extended polling Active polling Polling + piggy Active polling Active polling Figure 6.32 Mean signaling delay polling-based reservation MAC protocols rare requests (average packet size: 1500 bytes)
Tags :
signaling,extended,network,delay,protocols,load,piggybacking,area,requests,both,rare,basic,longer