STUDI MEKANISME KERJA TOKEN BUCKET UNTUK PEMBATAS KECEPATAN TRAFIK DATA

• Main Author: FW Hidayat, Bagus
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/4883
• Institution: Institut Teknologi Bandung

<b>Abstrarct:</b><p align="justify"> <br /> <br /> Committed access rate (CAR) service that limit data access rate on interface could be designed by a token bucket policing mechanism without buffer. jer. A token bucket mechanism with buffer is used to control delay or traffic shopping.<p align="justify"> <br /> This study was to analyse and simulate a standard token bucket mechanism and an extended token bucket mechanism. The packet's intererrival time was by exponential and Pareto distribution. The token bucket parameters behaviour were analysed by changing some input parameters. The token bucket parameters were: token rate r, bucket size b, extended bucket size eb, arrival packet rate R, conform packet rate .ice dan drop packet rate A<p align="justify"> <br /> The results of simulation and analysing were: the bigger bucket size b and the faster token arrival rate r, 2 would more conform to the specified rate. The extended token bucket mechanism produced more conform traffic than the standard token bucket mechanism. By small value r with the exponential traffic source, to produce conform traffic larger than 80% to 90%, we could set b 1 x r or b - 1.5 x r. By all value r, with the the standard taken bucket and the exponential traffic source, tb produce conform tragic larger than 92%, we could set b like the formula mentioned above or b equal to or 'larger than 6 times of average packet length. By b equal to 6 with the extended token bucket and the exponential traffic - source, the mechanism produced average conform traffic 99.3%. By b equal to 20 and 30 with extended token bucket and Pareto traffic source, it produced average conform traffic 96.8% and 98%. Conform traffic pattern of Pareto traffic source is less smooth than that of exponential traffic source.