Dahai Xu

Novel algorithms for shared segment protection

The major challenges in designing survivable schemes are how to allocate a minimal amount of spare resources (e.g., bandwidth) using fast (e.g., polynomial-time) algorithms, and, in case a failure occurs, to be able to recover quickly from it. All existing approaches invariably make tradeoffs. We propose novel shared segment protection algorithms which make little or no compromise . We develop an elegant integer linear programming (ILP) model to determine an optimal set of segments to protect a given active path. Although the ILP approach is useful for a medium-size network, it is too time consuming for large networks. Accordingly, we also design a fast heuristic algorithm based on dynamic programming to obtain a near-optimal set of segments. Although the heuristic algorithm has a polynomial time complexity, it can achieve a bandwidth efficiency as high as some best-performing shared path protection schemes and, at the same time, much faster recovery than these shared path protection schemes. The proposed scheme is also applicable to a wide range of networking technologies, including Internet Protocol and wavelength-division multiplexing networks under the generalized multiprotocol label switched framework.

Link-state routing with hop-by-hop forwarding can achieve optimal traffic engineering

This paper settles an open question with a positive answer: Optimal traffic engineering (or optimal multicommodity flow) can be realized using just link-state routing protocols with hop-by-hop forwarding. Todays typical versions of these protocols, Open Shortest Path First (OSPF) and Intermediate System-Intermediate System (IS-IS), split traffic evenly over shortest paths based on link weights. However, optimizing the link weights for OSPF/IS-IS to the offered traffic is a well-known NP-hard problem, and even the best setting of the weights can deviate significantly from an optimal distribution of the traffic. In this paper, we propose a new link-state routing protocol, PEFT, that splits traffic over multiple paths with an exponential penalty on longer paths. Unlike its predecessor, DEFT, our new protocol provably achieves optimal traffic engineering while retaining the simplicity of hop-by-hop forwarding. The new protocol also leads to a significant reduction in the time needed to compute the best link weights. Both the protocol and the computational methods are developed in a conceptual framework, called Network Entropy Maximization, that is used to identify the traffic distributions that are not only optimal, but also realizable by link-state routing.

Network Design and Architectures for Highly Dynamic Next-Generation IP-Over-Optical Long Distance Networks

The DARPA CORONET project seeks to develop the target network architectures and technologies needed to build next-generation long-distance IP-over-Optical-Layer (IP/OL) networks. These next-generation networks are expected to scale 10-100 times larger than todays largest commercial IP/OL network. Furthermore, DARPA has established advanced objectives for very rapid provisioning of new IP or private line connections, very rapid restoration against up to three simultaneous network failures, and future dynamic ldquowavelengthrdquo services ranging from speeds of 40-800 Gigabits per second. Besides these ambitious goals, the CORONET project seeks to establish a commercially-viable network architecture that supports both commercial and government services. In this paper, we describe the CORONET program requirements, and present our initial architectures and analysis of the early phases of this long-term project. We propose a novel 2-Phase Fast Reroute restoration method that achieves 50-100 ms restoration in the IP-Layer in a cost-effective manner, and a commercially viable OL restoration method that can meet the rapid CORONET requirements. We also estimate the magnitude of the extra capacity needed to provide dynamic wavelength services compared to that of static services, and show that the extra capacity to restore a small percentage of high priority traffic against multiple failures requires a small amount of extra capacity compared to that of single failures.

Architectures and Protocols for Capacity Efficient, Highly Dynamic and Highly Resilient Core Networks [Invited]

The Core Optical Networks (CORONET) program addresses the development of architectures, protocols, and network control and management to support the future advanced requirements of both commercial and government networks, with a focus on highly dynamic and highly resilient multi-terabit core networks. CORONET encompasses a global network supporting a combination of IP and wavelength services. Satisfying the aggressive requirements of the program required a comprehensive approach addressing connection setup, restoration, quality of service, network design, and nodal architecture. This paper addresses the major innovations developed in Phase 1 of the program by the team led by Telcordia and AT&T. The ultimate goal is to transfer the technology to commercial and government networks for deployment in the next few years.

Maximizing throughput for optical burst switching networks

In optical burst switching (OBS) networks, a key problem is to schedule as many bursts as possible on wavelength channels so that the throughput is maximized and the burst loss is minimized. Most of the current research on OBS has been concentrated on reducing burst loss in an ldquoaverage-caserdquo sense, and little effort has been devoted to understanding the worst case performance. Since OBS itself is an open-loop control system, it may exhibit a worst case behavior when adversely synchronized. On the other hand, most commercial systems require an acceptable worst case performance. In this paper, we use competitive analysis to analyze the worst case performance of a large set of scheduling algorithms, called best-effort online scheduling algorithms, for OBS networks and establish a number of interesting upper and lower bounds on the performance of such algorithms. Our analysis shows that the performance of any best-effort online algorithm is closely related to a few factors, such as the range of offset time, maximum-to-minimum burst-length ratio, and the number of data channels. A surprising discovery is that the worst case performance of any best-effort online scheduling algorithm is primarily determined by the maximum-to-minimum burst-length ratio, followed by the range of offset time. Furthermore, if all bursts have the same burst length and offset time, all best-effort online scheduling algorithms generate the same optimal solution, regardless of how different they may look. Our analysis can also be extended to some non-best-effort online scheduling algorithms, such as the well-known Horizon algorithm, and establish similar bounds. Based on the analytic results, we give guidelines for several widely discussed OBS problems, including burst assembly, offset time setting, and scheduling algorithm design, and propose a new channel reservation protocol called virtual fixed offset-time (VFO) to improve the worst case performance. Our simulation shows that VFO can also reduce the average burst loss rate.

Achieving fast and bandwidth-efficient shared-path protection

Dynamic provisioning of restorable bandwidth guaranteed paths is a challenge in the design of broad-band transport networks, especially next-generation optical networks. A common approach is called (failure-independent) path protection, whereby for every mission-critical active path to be established, a link (or node) disjoint backup path (BP) is also established. To optimize network resource utilization, shared path protection should be adopted, which often allows a new BP to share the bandwidth allocated to some existing BPs. However, it usually leads the backup paths to use too many links, with zero cost in term of additional backup bandwidth, along its route. It will violate the restoration time guarantee. In this paper, we propose novel integer linear programming (ILP) formulations by introducing two parameters (/spl epsi/ and /spl mu/) in both the sharing with complete information (SCI) scheme and the distributed partial information management (DPIM) scheme. Our results show that the proposed ILP formulations can not only improve the network resource utilization effectively, but also keep the BPs as short as possible.

DEFT: Distributed Exponentially-Weighted Flow Splitting

Network operators control the flow of traffic through their networks by adapting the configuration of the underlying routing protocols. For example, they tune the integer link weights that interior gateway protocols like OSPF and ISIS use to compute shortest paths. The resulting optimization problem -to find the best link weights for a given topology and traffic matrix -is computationally intractable even for the simplest objective functions, forcing the use of local-search techniques. The optimization problem is difficult in part because these protocols split traffic evenly along shortest paths, with no ability to adjust the splitting percentages or direct traffic on other paths. In this paper, we propose an extension to these protocols, called Distributed Exponentially-weighted Flow SpliTting (DEFT), where the routers can direct traffic on non-shortest paths, with an exponential penalty on longer paths. DEFT leads not only to an easier-to-solve optimization problem, but also to weight settings that provably perform no worse than OSPF and IS-IS. Furthermore, in our optimization problem, both link weights and flows of traffic are integrated as optimization variables into the formulation and jointly solved by a two-stage iterative method. Our novel formulation leads to a much more efficient way to identify good link weights than the local-search heuristics used for OSPF and IS-IS today. DEFT retains the simplicity of having routers compute paths based on configurable link weights, while approaching the performance of more complex routing protocols that can split traffic arbitrarily over any paths.

Link-State Routing with Hop-by-Hop Forwarding Can Achieve Optimal Traffic Engineering

Link-state routing with hop-by-hop forwarding is widely used in the Internet today. The current versions of these protocols, like OSPF, split traffic evenly over shortest paths based on link weights. However, optimizing the link weights for OSPF to the offered traffic is an NP-hard problem, and even the best setting of the weights can deviate significantly from an optimal distribution of the traffic. In this paper, we propose a new link-state routing protocol, PEFT, that splits traffic over multiple paths with an exponential penalty on longer paths. Unlike its predecessor, DEFT, our new protocol provably achieves optimal traffic engineering while retaining the simplicity of hop-by-hop forwarding. A gain of 15 % in capacity utilization over OSPF is demonstrated using the Abilene topology and traffic traces. The new protocol also leads to significant reduction in the time needed to compute the best link weights. Both the protocol and the computational methods are developed in a new conceptual framework, called network entropy maximization, which is used to identify the traffic distributions that are not only optimal but also realizable by link-state routing.

Elastic service availability: utility framework and optimal provisioning

Service availability is one of the most closely scrutinized metrics in offering network services. It is important to cost- effectively provision a managed and differentiated network with various service availability guarantees under a unified platform. In particular, demands for availability may be elastic and such elasticity can be leveraged to improve cost-effectiveness. In this paper, we establish the framework of provisioning elastic service availability through network utility maximization, and propose an optimal and distributed solution using differentiated failure recovery schemes. First, we develop a utility function with configurable parameters to represent the satisfaction perceived by a user upon service availability as well as its allowed source rate. Second, adopting Quality of Protection [1] and shared path protection, we transform optimal provisioning of elastic service availability into a convex optimization problem. The desirable service availability and source rate for each user can be achieved using a price-based distributed algorithm. Finally, we numerically show the tradeoff between the throughput and the service availability obtained by users in various network topologies. This investigation quantifies several engineering implications. For example, indiscriminately provisioning service availabilities for different kinds of users within one network leads to noteworthy sub-optimality in total network utility. The profile of bandwidth usage also illustrates that provisioning high service availability exclusively for critical applications leads to significant waste in bandwidth resource.

Secure Key Management Architecture Against Sensor-Node Fabrication Attacks

In lightweight mobile ad hoc networks, both probabilistic and deterministic key management schemes are fragile to node fabrication attacks. Our simulation results show that the Successful Attack Probability (SAP) can be as high as 42.6% with the fabrication of only 6 copies from captured nodes comprising only 3% of all nodes. In this paper, we propose two low-cost secure-architecture-based techniques to improve the security against such node fabrication attacks. Our new architectures, specifically targeted at the sensor-node platform, protect long-term keys using a root of trust embedded in the hardware System-on-a-Chip (SoC). This prevents an adversary from extracting these protected long-term keys from a captured node to fabricate new nodes. The extensive simulation results show that the proposed architecture can significantly decrease the SAP and increase the security level of key management for mobile ad hoc networks.