10.07.2015 Views

DOWNLOAD MY Ph.D Thesis - UNAM

DOWNLOAD MY Ph.D Thesis - UNAM

DOWNLOAD MY Ph.D Thesis - UNAM

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Chapter 7Performance optimisation for the support of TCIS pA) Performance analysis for the exponential backoff algorithmFrom simulation results, Figures 7.7 to 7.9 present a performance comparison betweenthe enhanced-reservation-request mechanisms when the exponential backoff algorithmis used.Results for mean access delay showed that on high traffic loads, above ≈ 33% of the cc(105 stations) for VoIP traffic (Figure 7.7a) and 46% of the cc (34 stations) for mixedtraffic (Figure 7.7b), all the enhanced mechanisms offered lower access delays than thepure reservation access mode. For both traffic types, it was found that the bestmechanism is the continuous piggyback request. This mechanism achieved a betterperformance because all requests are piggybacked on the same data messages. Also, ifthe buffer is empty when a station receives the piggybacked slots, such reserved slotsare then used to send further requests for at least one slot. Contention access is thereforeavoided. Thus, more bandwidth can be allocated to the reservation access region. Thisresults in the lowest access delays (during high traffic loads) and the highest systemthroughput.Mean Access Delay (ms)500450400350300250200150100503 Mbps Up.9.7 Kbps kbps VoIPBackoff CRAa) VoIP Traffic b) Mixed TrafficPRARRCRREPRAPGCPGPRARRCRRPG3 Mbps Up.Mixed TrafficBackoff CRAEPRACPG070 80 90 100 110 120 130 140 150 160 170umber of Active Stations31 33 35 37 39 41 43 45umber of Active StationsFigure 7.7 – Mean access delay vs. No. of active stations for VoIP and mixed traffic.Exponential backoff algorithm with enhanced-reservation-request mechanisms.7-8

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!