Polling, Greedy and Horizon Servers on a Circle

Service in a loop-based polling system consists of a single server moving around a closed tour, stopping to perform services wherever requests are encountered. There are N stations unit buffer queues spaced one unit of distance apart, and the server moves at a unit speed. All queues are identical, and the service time is deterministic. We compare the two well known cyclic polling and greedy servers with a new control policy called the horizon server. The cyclic polling server moves in one direction, even if no requests are waiting, and stops whenever it encounters a request. The greedy server selects the nearest request for its next service. At any station the greedy server can reverse its direction if a new request arrives nearby, and if no requests are waiting the greedy server does not move. The horizon server, with parameter d, ignores all requests for service from a distance farther than d,. Within its horizon ≤ d it acts like the greedy server. Analytical solutions for N = 2 and 3 and numerical results for N≤ 6 show that the horizon server, with the optimum value of d, outperforms the polling and the greedy servers.

[1]  Layek Abdel-Malek,et al.  The application of inverse kinematics in the optimum sequencing of robot tasks , 1990 .

[2]  Edward G. Coffman,et al.  Polling and greedy servers on a line , 1987, Queueing Syst. Theory Appl..

[3]  Nathan J. Muller,et al.  Internetworking Lans: Operation, Design and Management , 1992 .

[4]  Hideaki Takagi,et al.  Application of Polling Models to Computer Networks , 1991, Comput. Networks ISDN Syst..

[5]  Mischa Schwartz,et al.  Telecommunication networks: protocols, modeling and analysis , 1986 .

[6]  Moshe Sidi,et al.  Polling systems: applications, modeling, and optimization , 1990, IEEE Trans. Commun..

[7]  T.-H. Wu,et al.  SONET ring applications for survivable fiber loop networks , 1991, IEEE Communications Magazine.

[8]  Jeremiah F. Hayes.  Modeling and Analysis of Computer Communications Networks , 1984 .

[9]  Bezalel Gavish,et al.  Topological design of telecommunication networks-local access design methods , 1991, Ann. Oper. Res..

[10]  Hideaki Takagi,et al.  Analysis of polling systems , 1986 .

[11]  Bezalel Gavish,et al.  Topological design of centralized computer networks - formulations and algorithms , 1982, Networks.

[12]  Peter O'Reilly,et al.  Performance Analysis of Local Computer Networks , 1986 .

[13]  C. Mack,et al.  THE EFFICIENCY OF N MACHINES UNI-DIRECTIONALLY PATROLLED BY ONE OPERATIVE WHEN WALKING TIME AND REPAIR TIMES ARE CONSTANTS , 1957 .

[14]  Toby J. Teorey,et al.  A comparative analysis of disk scheduling policies , 1972, CACM.

[15]  Edward G. Coffman,et al.  A continuous polling system with constant service times , 1986, IEEE Trans. Inf. Theory.

[16]  Hideaki Takagi,et al.  Queuing analysis of polling models , 1988, CSUR.

[17]  Micha Hofri,et al.  On the Expected Performance of Scanning Disks , 1982, SIAM J. Comput..

[18]  C. Mack,et al.  Efficiency of Bi-Directionally Traversed Machines , 1973 .

[19]  Micha Hofri.  Disk scheduling: FCFS vs.SSTF revisited , 1980, CACM.