Burstification Queue Management in Optical Burst Switching Networks

Among the various issues lying in optical burst switching (OBS) networks, burstification, i.e., assembling multiple IP packets into bursts, is an important one. Between the two important aspects related to burst assembly, the burst assembly algorithm aspect has been extensively studied in the literature. However, as far as we know, there is no research about the burstification queue management (BQM) aspect, which refers to how many burstification queues (BQ) we should set at each OBS edge node and how to manage these BQs. Suppose there are G destinations (egress edge nodes) and the OBS network provides S different quality of service (QoS) classes. Traditionally, it is simply regarded that each ingress edge node needs G· S queues to sort incoming packets, one for each possible destination and QoS class. For simplicity, we call this policy the static dedicate BQM (SDB) policy. The SDB policy, though simple, lacks scalability since we have to add S BQs at each OBS edge node if an extra OBS edge node is added to the OBS network. To solve this problem, we propose in this paper two BQM policies: quasi-static BQM (QSB) policy and dynamic BQM (DB) policy. For the QSB policy, we derive the packet loss probability due to lacking BQs based on a Markov chain, from which we can work out the employed number of BQs for a given packet loss probability. Based on these results, the scalability of the QSB policy is also studied. With the DB policy, we not only can dynamically assign BQs for incoming packets, but also can dynamically allocate buffer capacity for each BQ by using a least-mean-square (LMS)-based linear prediction filter. The performance of the DB policy is investigated by analysis and extensive simulations. We also compared the performance of the QSB policy and the DB policy. Results from analysis and simulation demonstrate that the DB policy is the best.     

Using multiple per egress burstifiers for enhanced TCP performance in OBS networks

Burst assembly mechanism is one of the fundamental factors that determine the performance of an optical burst switching (OBS) network. In this paper, we investigate the influence of the number of burstifiers on TCP performance for an OBS network. The goodput of TCP flows between an ingress node and an egress node traveling through an optical network is studied as the number of assembly buffers per destination varies. First, the burst-length independent losses resulting from the contention in the core OBS network using a non-void-filling burst scheduling algorithm, e.g., Horizon, are studied. Then, burst-length dependent losses arising as a result of void-filling scheduling algorithms, e.g., LAUC-VF, are studied for two different TCP flow models: FTP-type long-lived flows and variable size short-lived flows. Simulation results show that for both types of scheduling algorithms, both types of TCP flow models, and different TCP versions (Reno, Newreno and Sack), TCP goodput increases as the number of burst assemblers per egress node is increased for an OBS network employing timer-based assembly algorithm. The improvement from one burstifier to moderate number of burst assemblers is significant (15–50% depending on the burst loss probability, per-hop processing delay, and the TCP version), but the goodput difference between moderate number of buffers and per-flow aggregation is relatively small, implying that an OBS edge switch should use moderate number of assembly buffers per destination for enhanced TCP performance without substantially increasing the hardware complexity.