Wojciech Jawor

Improved online algorithms for buffer management in QoS switches

We consider the following buffer management problem arising in QoS networks: Packets with specified weights and deadlines arrive at a network switch and need to be forwarded so that the total weight of forwarded packets is maximized. Packets not forwarded before their deadlines are lost. The main result of the article is an online 64/33 ≈ 1.939-competitive algorithm, the first deterministic algorithm for this problem with competitive ratio below 2. For the 2-uniform case we give an algorithm with ratio ≈ 1.377 and a matching lower bound.

A Note on Scheduling Equal-Length Jobs to Maximize Throughput

The main aim of this note is to show that a polynomial-time algorithm for the scheduling problem 1|r j ; p j = p| ∑ Uj given by Carlier in (1981) is incorrect. In this problem we are given n jobs with release times and deadlines. All jobs have the same processing time p. The objective is to find a non-preemptive schedule that maximizes the number of jobs completed by their deadlines. The feasibility version of this problem, where we ask whether all jobs can meet their deadlines, has been studied thoroughly. Polynomial-time algorithms for this version were first found, independently, by Simons (1978) and Carlier (1981). A faster algorithm, with running time O(n log n), was subsequently given by Garey et al. (1981). The elegant feasibility algorithm of Carlier (1981) is based on dynamic programming and it processes jobs from left to right on the time-axis. For each time t, it constructs a partial schedule with jobs that complete at or before time t. Carlier also attempted to apply the same technique to design a polynomial-time algorithm for the maximization version, 1|r j ; p j = p| ∑ Uj , and claimed an O(n3 log n)-time algorithm. His result is now widely cited in the literature. We show, however, that this algorithm is not correct, by giving an instance on which it produces a sub-optimal schedule. Our counter-example can be, in fact, extended to support a broader claim, namely that even the general approach from Carlier (1981) cannot yield a polynomial-time algorithm. By this general approach we mean a class of algorithms that processes the input from left to right and make decisions based on the deadline ordering, and not their exact values. The question remains as to how efficiently can we solve the scheduling problem 1|r j ; p j = p| ∑ Uj . Baptiste (1999) gave an O(n7)-time algorithm for the more general version of this problem where jobs have weights. We show how to modify his algorithm to obtain a faster, O(n5)-time algorithm for the non-weighted case. These last two results are discussed only briefly in this note. The complete proofs can be found in the full version of this paper, see Chrobak et al. (2004).

Online Competitive Algorithms for Maximizing Weighted Throughput of Unit Jobs

We study an online scheduling problem for unit-length jobs, where each job is specified by its release time, deadline, and a nonnegative weight. The goal is to maximize the weighted throughput, that is the total weight of scheduled jobs. We first give a randomized algorithm RMix with competitive ratio of e/(e-1)≈ 1.582. Then we consider s-bounded instances where the span of each job is at most s. We give a 1.25-competitive randomized algorithm for 2-bounded instances, and a deterministic algorithm Edf α , whose competitive ratio on s-bounded instances is at most 2-2/s+o(1/s). For 3-bounded instances its ratio is φ ≈ 1.618, matching the lower bound.

Preemptive online scheduling: optimal algorithms for all speeds

Our main result is an optimal online algorithm for preemptive scheduling on uniformly related machines with the objective to minimize makespan. The algorithm is deterministic, yet it is optimal even among all randomized algorithms. In addition, it is optimal for any fixed combination of speeds of the machines, and thus our results subsume all the previous work on various special cases. Together with a new lower bound it follows that the overall competitive ratio of this optimal algorithm is between 2.054 and e≈ 2.718.

Improved Online Algorithms for Buffer Management in QoS Switches

The following buffer management problem is studied: packets with specified weights and deadlines arrive at a network switch and need to be forwarded so that the total weight of forwarded packets is maximized. A packet not forwarded before its deadline brings no profit. The main result is an online \(\frac{64}{33}\approx 1.939\)-competitive algorithm, the first deterministic algorithm for this problem with competitive ratio below 2. In the s-uniform case, where for all packets the deadline equals the arrival time plus s, we give an \({5}-\sqrt{10} \approx 1.838\)-competitive algorithm. This algorithm achieves the same ratio in a more general scenario when all packets are similarly ordered. For the 2-uniform case we give an algorithm with ratio ≈ 1.377 and a matching lower bound.

Online Scheduling of Equal-Length Jobs: Randomization and Restarts Help

The input of the studied scheduling problem is a set of jobs with equal processing times, where each job is specified by its release time and deadline. The goal is to determine a single-processor, non-preemptive schedule that maximizes the number of completed jobs. In the online version, each job arrives at its release time.

Preemptive Online Scheduling: Optimal Algorithms for All Speeds

Our main result is an optimal online algorithm for preemptive scheduling on uniformly related machines with the objective to minimize makespan. The algorithm is deterministic, yet it is optimal even among all randomized algorithms. In addition, it is optimal for any fixed combination of speeds of the machines, and thus our results subsume all the previous work on various special cases. Together with a new lower bound it follows that the overall competitive ratio of this optimal algorithm is between 2.054 and e≈2.718. We also give a complete analysis of the competitive ratio for three machines.

Three dozen papers on online algorithms

This column contains a summary of last years research on online algorithms presented at the STOC, FOCS, ICALP, ESA, and STACS conferences. Unfortunately, due to space constraints, the report could not be entirely exhaustive, and results from other conferences or journal articles are not covered. We hope that all readers will find in the survey something of interest, to fill those long winter evenings. The papers in the report are organized roughly by applications.

Competitive Analysis of Scheduling Algorithms for Aggregated Links

Abstract We study an online job scheduling problem arising in networks with aggregated links. The goal is to schedule n jobs, divided into k disjoint chains, on m identical machines, without preemption, so that the jobs within each chain complete in the order of release times and the maximum flow time is minimized. We present a deterministic online algorithm

Experimental Analysis of Scheduling Algorithms for Aggregated Links

\mathsf{Block}