A statistical network service which allows a certain fraction of traffic to not meet its QoS guarantees can extract additional capacity from a network by exploiting the statistical properties of the traffic. Here we consider a statistical service which assumes statistical independence of flows, but does not make any assumptions on the statistics of traffic sources, other than that they are regulated, e.g., by a leaky bucket. Under these conditions, we present functions, so-called local effective envelopes and global effective envelopes, which are, with high certainty, upper bounds of multiplexed traffic. We show that these envelopes can be used to obtain bounds on the amount of traffic on a link that can be provisioned with statistical QoS. A key advantage of our bounds is that they can be applied with a variety of scheduling algorithms. In fact, we show that one can reuse existing admission control functions that are available for scheduling algorithms with a deterministic service. We present numerical examples which compare the number of flows with statistical QoS guarantees that can be admitted with our effective envelope approach to those achieved with existing methods.

Network services for the most demanding advanced networked applications which require absolute, per-flow service assurances can be deterministic or statistical. By exploiting the statistical properties of traffic, statistical assurances can extract more capacity from a network than deterministic assurances. We consider statistical service assurances for traffic scheduling algorithms. We present functions, so-called effective envelopes, which are, with high certainty, upper bounds of multiplexed traffic. Effective envelopes can be used to obtain bounds on the amount of traffic on a link that can be provisioned with statistical service assurances. We show that our bounds can be applied to a variety of traffic scheduling algorithms. In fact, one can reuse existing admission control functions for scheduling algorithms with deterministic assurances. We present numerical examples which compare the number of flows with statistical assurances that can be admitted with our effective envelope approach to those achieved with existing methods.

Investigates the problem of management and control in a large and, for simplicity, homogeneous packet-switched network. Specifically the authors focus on routing, an important function of network management. The network consists of several individually controlled domains. Domains are interconnected via gateway links. Each domain is controlled by its own network control center, while the overall network performance is managed by an integrated network control center. Each center has only a portion of the information required for global routing. The authors investigate the impact of the reduced information available at each center on network performance (average delay in the present case). They present a general approach to designing a hierarchical algorithm for routing in multidomain networks. They propose a heuristic procedure suitable for packet-switched networks. Several numerical examples illustrate the impact of incomplete information on the network. Performance is compared with a lower bound obtained, which is not differentiating destinations in other domains. Therefore, for this bound, each domain is perceived as a single node in a simplified model of the network.>

The problem of management and control in a large and, for simplicity, homogeneous packet-switched network is investigated. In particular, routing is considered as an important function of network management. The network consists of several individually controlled domains interconnected via gateway links. Each domain is controlled by its own network control center, while the overall network performance is managed by the integrated network control center. Each center has only a portion of the information required for global routing. The impact of the reduced information available at each center on network performance is investigated. A general approach to designing a hierarchical algorithm for routing in multi-domain networks is presented. A heuristic procedure suitable for packet-switched networks is proposed. An example of the algorithm is shown. Its performance is compared against a lower bound to the network performance.>

An investigation is conducted of routing in packet-switched networks that consist of two individually controlled domains. Each domain controller determines a static routing plan for requirements that are contained within its domain and for requirements that are addressed to nodes in the other domain. The authors propose two routing algorithms, assuming that each controller has limited information on the configuration of the other domain. Numerical examples are provided that compare the delay performance of such algorithms with the optimal routing of a centralized scheme that has full information and control over both domains. Implications of distributed control on network design are explored.>

The problem of analyzing the thoughput of packet radio networks with realistic topologies is considered. We present an algorithm for the solution of this problem and show that both the memory requirements and running time of this algorithm in practice grow polynomially with the size of the problem. Although in theory both can grow exponentially in the worst case, we offer computational experience with the procedure and show that for realistic topologies where connectivity is related to distance, the rate of growth is quadratic in the number of links. Even for regular grids, which are pathological in their symmetry, the rate of growth is only cubic in the number of links. We thus conclude that the procedure is effective for realistic topologies with up to several hundred nodes.

In this paper, we use a Markov model to develop a product form solution to efficiently analyze the throughput of arbitrary topology multihop packet radio networks that employ a carrier sensing multiple access (CSMA) protocol with perfect capture. We consider both exponential and nonexponential packet length distributions. Our method preserves the dependence between nodes, characteristic of CSMA, and determines the joint probability that nodes are transmitting. The product form analysis provides the basis for an automated algorithm that determines the maximum throughput in networks of size up to 100 radio nodes. Numerical examples for several networks are presented. This model has led to many theoretical and practical extensions. These include determination of conditions for product form analysis to hold, extension to other access protocols, and consideration of acknowledgments.

Abstract The problem considered is that of finding an optimal (minimum cost) design for a centralized processing network given a set of locations, traffic magnitudes between these locations, and a single common source or destination. Several heuristics, which are efficient (in terms of their execution time and memory requirements on a digital computer) and which produce seemingly good results, have already been developed and are currently accepted techniques. Some work has also been done on finding optimal solutions to this problem both as a design tool and as a means of verifying the effectiveness of proposed heuristics. We focus on this latter area. Currently known techniques for the optimal solution of this problem via integer programming have fallen short of the desired objectives as they require too much memory and running time to be able to treat problems of realistic size and complexity. We develop an improved technique which is capable of handling more realistic problems.

A two-level adaptive routing scheme for packet-switched computer communication networks is proposed and investigated. The first level is quasi-static and based on the global network status. The second level is dynamic with decisions being made at each node in an attempt to obtain the savings in average delay predicted by a multiserver model of the node. Simulations confirm the predicted improvement.

A basic property of data sources in interactive applications is burstiness, i.e., short periods of activity followed by long idle periods. In these same applications message delay is the primary performance criterion. The combination of bursty flow and a delay criterion leads to a source encoding problem in which delay plays a central role. A salient feature of this problem is that there is a tradeoff between delay and the number of protocol bits required to represent the state of the source. A simple model of a bursty source is studied with the objective of understanding the relationship between coding efficiency and delay. Two encoding schemes, a block encoding technique and a technique employing flags, are examined in some detail. For both the block encoding and the flag schemes, a significant result is that as the source becomes less bursty, delay grows without bound. This result is obtained in spite of the fact that both schemes are reasonable and in the limit the encoding problem disappears. It also appears that the flag encoding technique has much smaller delay than block encoding.

We consider the problem of designing a centralized telecommunication network comprised of multipoint lines given a set of terminal locations, traffic requirements, and a common central site. The optimal solution to this problem is a capacitated minimal spanning tree. We develop a class of heuristic algorithms for the solution of this problem by imbedding existing heuristics, referred to as first-order greedy algorithms, inside a loop where small, carefully chosen sets of arcs are alternately forced in and out of the solution. The resultant procedure is shown to be superior to existing techniques, producing solutions typically 2 percent better, while requiring only a modest amount of additional computer time.

This correspondence examines multiparameter generalizations of the Cramér-Rao (C-R) bound and related bounds from a new viewpoint. We derive a general class of bounds and show that Rao's generalization is the tightest (bes0 of the class. A bound reported by Zacks is another member of the class. This derivation of the C-R bound emphasizes its optimum nature. The relationship of the general class to Barankin bounds is also discussed.

Estimation of the parameters of a single-frequency complex tone from a finite number of noisy discrete-time observations is discussed. The appropriate Cramér-Rao bounds and maximum-likelihood (MI.) estimation algorithms are derived. Some properties of the ML estimators are proved. The relationship of ML estimation to the discrete Fourier transform is exploited to obtain practical algorithms. The threshold effect of one algorithm is analyzed and compared to simulation results. Other simulation results verify other aspects of the analysis.

The classical problem of detecting known signals in additive nonstationary nonwhite noise is treated using state-variable methods. The noise is assumed to be generated as the solution of a linear time-varying differential equation driven by white noise. The Markov property of the noise is used to derive a difference equation for the likelihood ratio. The terms in the difference equation are given directly in terms of the coefficients of the differential equation defining the noise. These results, being in recursive form, reduce considerably the computational effort when discrete samples are used. The signal-to-noise ratio is also derived. Explicit expressions for the likelihood ratio are also obtained for continuous processing. These results are given directly in terms of the signals and the noise-generating equation. Thus, there are no integral equations to solve. The signal-to-noise ratio is calculated explicitly for both stationary and nonstationary noise.