Kevin W. Lu

A framework for characterizing disaster-based network survivability

This paper formulates a general framework that includes and extends the existing definitions for network survivability. Based on this framework, network survivability is characterized by a survivability function rather than a single-value survivability measure, and various quantities of interest can be derived from the function. Examples are the expected survivability, the worst-case survivability, the r-percentile survivability, and the probability of zero survivability. The survivability function is especially useful for the study of large-scale disasters. For illustration, the authors derive the survivability function in closed form for a simple ring network under link failures. They also discuss the general procedure for finding survivability functions for complex networks, and show that the survivability function reveals useful information about a network. This framework provides a unified and practical approach to analyzing and designing highly survivable communications networks. >

Comparison of buffering strategies for asymmetric packet switch modules

The performance of a class of asymmetric packet switch modules with channel grouping is analyzed. The switch module considered has n inputs and m outputs. A packet destined for a particular output address (out of g) needs to access only one of the r available physical output ports: m=gr. These switch modules are the key building blocks in many large multistage switch architectures. The focus is on the performance of input-buffered and output-buffered switch modules under geometrically bursty traffic. A combination of exact derivation, numerical analysis, and simulation yields the saturation throughput of input-buffered switch modules and the mean delay of the input-buffered and output-buffered switch modules. Tables and formulas useful for traffic engineering are presented. The results show that increasing the number of output ports per output address (r) can significantly improve switch performance, especially when traffic is bursty. Although output-buffered switch modules have significantly better performance than input-buffered switch modules when there are equal numbers of input and output ports, this performance difference becomes significantly smaller when the switch dimensions are asymmetric. >

A 3-stage interconnection structure for very large packet switches

Proposes a three-stage broadband packet-switch architecture for a future central office with more than 16000 ports. The switch is constructed by interconnecting many small independent switch modules, which can be implemented using modifications of various well-studied switch fabric designs. Multiple paths are provided for each input-output pair, and the channel grouping technique is used to decrease delay and increase throughput. A datagram packet routing approach is adopted in order to eliminate the table lookup that will be required by virtual-circuit routing. Ways of guaranteeing the sequence integrity of packets are discussed. It is estimated from performance analyses that 32768-port switches can be constructed and can perform well based on switch fabrics of no more than 128 ports.<<ETX>>

System and cost analyses of broad-band fiber loop architectures

Analyses of four broadband fiber-optic subscriber loop architectures, including active (high-speed time division multiplexing (TDM)-based) and passive (dense wavelength division multiplexing (WDM)-based, WDM-based with an analog subcarrier-multiplexing overlay, and splitter-based) double-star topologies, are presented. The analyses focus on specific demonstrated architectures and use component cost projections based on learning curves to estimate future network costs on a per-subscriber basis. Also investigated is the sensitivity of projected cost-per-subscriber to remote multiplexing node size and to double-star prove-in distance. The results indicate that the four architectures have very different double-star prove-in distances and that loop costs are minimized for much smaller remote node sizes than active loops, thus permitting cost-effective deployment of passive loops for smaller groups of subscribers. In addition, cost breakdowns for the four architectures indicate that splitter-based passive loops share electronics more effectively among subscribers than loop architectures requiring dedicated (per-subscriber) electronic interfaces, resulting in projected cost advantages for the splitter-based networks. >

A framework for network survivability characterization

The authors formulate a general framework that includes and extends the existing definitions for network survivability. Based on this framework, network survivability is characterized by a survivability function rather than a single-value survivability measure, and various quantities of interest can be derived from the function. Examples are the expected survivability, the worst-case survivability, the r-percentile survivability, and the probability of zero survivability. The survivability function is especially useful for the study of large-scale disasters. For illustration, the authors derive the survivability function in closed form for a simple ring network under link failures. They also discuss the general procedure for finding survivability functions for complex networks, and show that the survivability function reveals useful information about a network. This framework provides a unified and practical approach to analyzing and designing highly survivable communications networks.<<ETX>>

Performance analysis of asymmetric packet switch modules with channel grouping

The switch modules are studied because they are the key building blocks in large multistage switch architectures. The switch module considered has n inputs and m outputs. A packet destined for a particular output address (out of g) needs to access only one of the r available physical output ports: m=gr. Input-buffered, output-buffered, and unbuffered switch modules are studied. The results show that increasing the number of output ports per output address (r) can significantly improve the performance of buffered as well as unbuffered switch modules. For acceptable performance, the difference in throughput between buffered and unbuffered switch modules is considerable. For buffered switch modules, an interesting observation is that although output-buffered switch modules have significantly better delay performance than input-buffered switch modules when n=gr, the performance difference is diminished as one deviates from these switch dimensions.<<ETX>>

Cost analyses of emerging broadband fiber loop architectures

Cost analyses and comparisons of active and passive broadband fiber optic subscriber loop architectures, including all network components from the customer premises interface to the central office switch termination are presented. The analyses focus on specific, demonstrated architectures with projected availabilities in the early to mid-1990s. The installed first costs per subscriber for active and passive architectures are shown to be comparable. The passive loop, however, achieves this cost at a much smaller remote mode size than the active loop, implying cost-effective deployment of the passive loop for smaller groups of subscribers. In addition, key cost elements and tradeoffs for the architectures are identified, and sensitivities of double-star prove-in distances and network costs to laser, wavelength-division multiplexer (WDM), and fiber costs are examined. Finally, incremental network costs are compared for various business and residential service capabilities, and it is shown that passive loops have lower incremental costs for switched video.<<ETX>>

Comparative video distribution techniques for broadband fiber-optic subscriber loops

The authors compare a variety of possible video distribution techniques for future broadband fiber-optic subscriber loops. The analysis encompasses several families of passive-loop architectures recently proposed in the literature, including both dense WDM (wavelength division multiplexing)-based and splitter-based designs. For these architectures, technological and economic tradeoffs between switched and broadcast transport of video are examined as a function of various system parameters, using component cost projections for the middle to late 1990s. These system issues are investigated both for conventional video (NTSC-quality) and for advanced television services, such as HDTV (high-definition TV). For splitter-based architectures, the study predicts roughly equal costs for switched and broadcast delivery of video. For the WDM-based designs, a cost advantage of 10% to 20% is found for switched delivery. Cost increments as high as 25% or more are projected for upgrading the splitter-based systems from NTSC-quality video to a full menu of HDTV channels.<<ETX>>

Evolution of fiber-to-the-curb networks toward broadband capabilities

Proposals for fiber-to-the-curb (FTTC) systems have been made in recent years as optical system researchers and telecommunication providers have sought ways to deploy optical technologies close to the home based only on plain-old-telephone service (POTS) revenues. Fiber-to-the-home (FTTH) systems in contrast offer maximum bandwidth per subscriber but at significantly greater cost per subscriber today. One issue to explore with FTTC systems is their ability to support future broadband services since FTTC systems do not bring fiber all the way to customer premises. This paper presents an economic and technical analysis of various FTFC upgrade alternatives from a system supporting only POTS to later-generation systems also capable of broadcast video transport.

Analysis and applications of r-for N protection system

The availability of a single member of a group of N functionally identical systems that work in parallel and share a pool of r backup systems is analyzed. These N+r systems, each with failure rate lambda , are repairable, and there are m repairers available, each operating at rate mu . A Markov model is used to derive formulas for availability, mean time to failure, and mean time to repair for such a system. The analysis results indicate that by increasing r and decreasing rho , the ratio of failure rate to repair rate, overall network availability is improved. It is shown that, for certain rho s, some r:N pooled protections can significantly outperform 1:1 or 1:2 dedicated protection in terms of availability. The application of r:N protection to high-capacity transport systems is discussed, and it is indicated that multiwavelength systems may use additional wavelengths for protection. It is concluded that r:N protection can potentially provide both improved availability and significant cost advantages to the high-capacity transport systems.<<ETX>>