Amortized efficiency of list update and paging rules
Communications of the ACM
Online computation and competitive analysis
Online computation and competitive analysis
Competitve buffer management for shared-memory switches
Proceedings of the thirteenth annual ACM symposium on Parallel algorithms and architectures
Competitive queueing policies for QoS switches
SODA '03 Proceedings of the fourteenth annual ACM-SIAM symposium on Discrete algorithms
Nearly optimal FIFO buffer management for two packet classes
Computer Networks: The International Journal of Computer and Telecommunications Networking
Buffer Overflow Management in QoS Switches
SIAM Journal on Computing
The zero-one principle for switching networks
STOC '04 Proceedings of the thirty-sixth annual ACM symposium on Theory of computing
Harmonic buffer management policy for shared memory switches
Theoretical Computer Science - Special issue: Online algorithms in memoriam, Steve Seiden
Randomized qeue management for DiffServ
Proceedings of the seventeenth annual ACM symposium on Parallelism in algorithms and architectures
Competitive queue policies for differentiated services
Journal of Algorithms
On the Performance of Greedy Algorithms in Packet Buffering
SIAM Journal on Computing
Improved Lower Bounds for Competitive Ratio of Multi-Queue Switches in QoS Networks*
IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences
Scheduling policies for CIOQ switches
Journal of Algorithms
An improved algorithm for CIOQ switches
ACM Transactions on Algorithms (TALG)
IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences
A tight bound on online buffer management for two-port shared-memory switches
Proceedings of the nineteenth annual ACM symposium on Parallel algorithms and architectures
Lower and upper bounds on FIFO buffer management in QoS switches
ESA'06 Proceedings of the 14th conference on Annual European Symposium - Volume 14
Maximizing throughput in multi-queue switches
Algorithmica
Packet mode and QoS algorithms for buffered crossbar switches with FIFO queuing
Proceedings of the twenty-seventh ACM symposium on Principles of distributed computing
Packet buffering: randomization beats deterministic algorithms
STACS'05 Proceedings of the 22nd annual conference on Theoretical Aspects of Computer Science
A survey of buffer management policies for packet switches
ACM SIGACT News
Hi-index | 0.00 |
The online buffer management problem formulates the problem of queuing policies of network switches supporting QoS (Quality of Service) guarantee. We focus on multi-queue switches in QoS networks proposed by Azar et al. They introduced so-called "the relaxed model". Also, they showed that if the competitive ratio of the single-queue model is at most c, and if the competitive ratio of the relaxed model is at most c2, then the competitive ratio of the multi-queue switch model is cc2. They proved that c2d2, and obtained upper bounds on the competitive ratios for several multi-queue switch models. In this paper, we propose an online algorithm called DS (Dual Scheduling) for the competitive ratio of the relaxed model and obtain some better competitive ratios of the 2-value multi-queue switch model, where the value of packets is restricted to 1 and α(e1). DS uses as subroutine any online algorithms A for the non-preemptive unit-value switch model, which has also been extensively studied. We prove that if the competitive ratio of A is at most c, then the competitive ratio of DS is at most αc(2--c) + c2--2c+2Dα(2--c)+c--1, which is strictly better than 2. The followings are a couple of examples of the improvement on the competitive ratios of the 2-value multi-queue switch models using our result: (i) We have improved the competitive ratio of deterministic algorithms for the non-preemptive 2-value multi-queue switch model from 4 to 3.177 for large enough B, where B is the number of packets each queue can simultaneously store. (ii) We have proved that the competitive ratio of randomized algorithms for the non-preemptive 2-value multi-queue switch model is at most 17D2--30≈3.023 for large enough B.