Efficient resource management for hard real-time communication over differentiated services architectures

Abstract

We propose an efficient strategy for resource management for scalable QoS guaranteed real-time communication services. This strategy is based on sink trees, and is particularly well suited for differentiated-services based architectures. We first show that finding a set of sink-trees in a given network is NP-complete. Then we propose a heuristic algorithm that always efficiently produces a set of sink-trees for a given network. Sink-tree based resource management integrates routing and resource reservation along the routes, and therefore has a number of advantages over other resource management scheme, in terms of: admission probability, link resource utilization, flow set up latency, signaling overhead, and routing overhead. In this paper we show by simulation experiments that even for simple cases the sinktree based approach shows excellent results in terms of admission probability.

A novel approach to scalable CAC for real-time traffic in sink-tree networks with aggregate scheduling

“…As a consequence, the following three admission control tests should be performed: 1. Check if there is sufficient bandwidth and buffer to accept the candidate flow along the path, i.e. if (4) and (12) would still hold after admitting the flow.…”

“…On the other hand, Multi-Protocol Label Switching (MPLS, [2]), that allows one to classify flows into forwarding equivalence classes (FECs) and to perform forwarding and routing on a per-FEC basis, is currently employed for supporting advanced traffic engineering schemes. Some recent work has focused on employing a sink-tree resource management scheme in both DiffServ ( [3], [4]) and MPLS ( [5], [6], [7], [8]) networks. Sink-tree resource management consists in partitioning a network in a set of logical trees, rooted at egress nodes, so that traffic directed towards an egress node is routed through the related tree.…”

“…For instance, in MPLS, this is done by establishing multipoint-to-point Label Switched Paths, which also allows for saving label space compared to a point-to-point strategy, thus achieving better scalability. How to efficiently establish an optimal set of sink trees in networks of arbitrary topology has also been the focus of recent researches [4], [7], [8]. Hereafter, we call sink-tree network (also called accumulation or multipoint-topoint) one in which a sink-tree resource management scheme is in place.…”

“…Note that any algorithm that is based on checking the above three conditions is optimum, because it only rejects a candidate flow if a chance actually exists that accepting it would make a bit of its (or some other already established flow's) traffic violate its required delay bound. Moreover it allows both bandwidth and buffer to be utilized up to 100% (see (4) and (12)). Besides testing if a candidate flow can be admitted, an admission control algorithm has to update the state of the network (e.g.…”

“…Consider the example shown in Fig. 3, in which three flows 1, 2 and 4 are established in the sink tree along paths 1 4 B includes all nodes in path 4 P . Now, from (6), after the admittance of a new flow the rate of interfering flows at node 5 can only increase.…”