DocumentCode
2475068
Title
Scheduling communication networks carrying real-time traffic
Author
Lehoczky, John P.
Author_Institution
Dept. of Stat., Carnegie Mellon Univ., Pittsburgh, PA, USA
fYear
1998
fDate
2-4 Dec 1998
Firstpage
470
Lastpage
479
Abstract
The paper presents new methods to analyze the performance of certain scheduling algorithms for communication networks carrying real time traffic and operating in a packet switched mode. The real time traffic consists of distinct sessions, each of which transmits packets with hard deadlines. Edf and fifo scheduling policies are analyzed in detail. For the single node case, methods based on real time queueing theory and heavy traffic queueing theory are applied to develop closed form expressions for the fraction of packets which miss their deadline as a function of the characteristics of the sessions using the node. The results are extended to a two node non Jackson network. Simulations illustrate the excellent accuracy of the closed form expressions. The formulas developed are used to design system parameters that can attain desired levels of packet lateness
Keywords
packet switching; queueing theory; real-time systems; scheduling; telecommunication computing; telecommunication traffic; closed form expressions; communication network scheduling; distinct sessions; edf scheduling; fifo scheduling policies; hard deadlines; heavy traffic queueing theory; packet lateness; packet switched mode; real time queueing theory; real time traffic; scheduling algorithm performance; single node case; system parameters; two node non Jackson network; Algorithm design and analysis; Communication networks; Communication switching; Packet switching; Performance analysis; Queueing analysis; Scheduling algorithm; Telecommunication traffic; Traffic control;
fLanguage
English
Publisher
ieee
Conference_Titel
Real-Time Systems Symposium, 1998. Proceedings., The 19th IEEE
Conference_Location
Madrid
Print_ISBN
0-8186-9212-X
Type
conf
DOI
10.1109/REAL.1998.739780
Filename
739780
Link To Document