Martin I. Eiger

Network restoration under a single link or node failure using Preconfigured Virtual Cycles

This paper presents a design algorithm for networks with a restoration mechanism that provides failure-independent, end-to-end path protection to a set of given demands under a single link or node failure with a focus on optical networks. The restoration routes are provided on preconfigured cycles, where each of the demands is assigned a single restoration route and specific restoration wavelengths on a segment of one cycle (splitting is not allowed). The number of reserved restoration wavelengths may vary from one link to the next on a cycle; hence, we refer to these cycles as Preconfigured Virtual Cycles (PVCs). The network design algorithm consists of three major parts. The first part generates a large number of candidate PVCs. Our algorithm allows assignment of certain demands that have common failure scenarios to the same PVC. The second part selects a set of PVCs from among the candidates, attempting to minimize the total reserved restoration cost while ensuring that each demand is assigned to one PVC. This is achieved by solving a set covering problem followed by elimination of duplicate assignments. The third part resolves conflicts of wavelength assignments.

Network restoration under dual failures using path-protecting preconfigured cycles

This paper presents a design algorithm for networks with a restoration mechanism that provides end-to-end path protection to a set of demands with specified working routes, where demands must survive one or two failures in optical networks and these failures may occur almost instantaneously. Each of the demands protected from two failures is assigned two restoration routes and restoration wavelengths on a cycle. A demand protected from a single failure is assigned one restoration route and restoration wavelengths on a cycle. These assignments are preconfigured so that switching and wavelength conversions are not needed at intermediate nodes of restoration routes. Splitting of demand across multiple restoration routes upon a failure occurrence is not allowed. The algorithm generates a large number of candidate Path-Protecting Preconfigured Cycles (PP-PCs). A candidate cycle may provide protection to a mix of demands requiring different protection levels, where demands may share under certain conditions restoration wavelengths even if they are exposed to common failure scenarios and have overlapping restoration routes. A near-optimal set of preconfigured cycles is selected from among all candidates, attempting to minimize the total cost of restoration wavelengths while ensuring that each demand is assigned to a single preconfigured cycle. This is achieved by solving a set covering problem followed by deleting duplicate demand assignments and by resolving wavelength assignment conflicts.

Node Placement and Sizing for Copper Broadband Access Networks

We consider a node placement and sizing problem that arises in certain types of broadband access architectures, such as ADSL and FTTC. We consider three variants of the problem that become progressively more restrictive, to capture realistic planning concerns and to produce solutions that have desirable practical attributes. A distinguishing feature of the problem is a constraint that limits the distance between each customer and the placed node that is assigned to serve it. We present a dynamic programming algorithm to solve the two most practical variants of the problem, and we provide computational results for both realistic and randomly generated test problems.

Client-Server Caching with Expiration Timestamps

We study client-server caching of data with expiration timestamps. Although motivated by the potential for caching in telecommunication applications, our work extends to the general case of caching data that has known expiration times. Toward this end, we tailor caching algorithms to consider expiration timestamps. Next, we consider several different client-server paradigms that differ in whether and how the server updates client caches. Finally, we perform simulation studies to evaluate the empirical performance of a variety of strategies for managing a single cache independent of the server and for managing caches in a client-server setting.

The effect of packetization on voice capacity in IEEE 802.11b networks

This paper reports on simulation studies of voice services in a wireless local area network (WLAN). We focus on the widespread IEEE 802.11b standard and present both analytical models and simulation results. Using quantifiable metrics based on packet losses and delays, we advance a definition of access point (AP) capacity as the maximum number of simultaneous voice calls that an AP can support with a desired level of service quality. We examine the role of codec packetization on WLAN channel efficiency, and through simulation show its significant impact on voice capacity. We further report on the dynamics of the collision probability, packet drop rate and medium activity time as more active voice users enter the system.

The Zodiac Policy Subsystem: A Policy-Based Management System for a High-Security MANET

Zodiac (Zero Outage Dynamic Intrinsically Assurable Communities) is an implementation of a high-security MANET, resistant to multiple types of attacks, including Byzantine faults. The Zodiac architecture poses a set of unique system security, performance, and usability requirements to its policy-based management system (PBMS). In this paper, we identify theses requirements, and present the design and implementation of the Zodiac Policy Subsystem (ZPS), which allows administrators to securely specify, distribute and evaluate network control and system security policies to customize Zodiac behaviors. ZPS uses the Keynote language for specifying all authorization policies with simple extension to support obligation policies.

Data caching for telephony services

We study client-server data caching, both with and without expiration timestamps, to assess its applicability to present and future telecommunications services such as local number portability, toll-free numbers, and mobile telephony. We perform simulation studies to evaluate the empirical performance of a variety of strategies for caching in a client-server setting. Data caching at client locations is found to be an economical and scalable approach to support future data-intensive telecommunications services.

Caching for Mobile Communication

We study caching as a means to reduce the message traffic and database accesses required for locating called subscribers in a Personal Communication Services (PCS) network. The challenge of caching routing information for mobile clients lies in the uncertainty of the length of time that the information remains valid.We use expiration timestamps to safeguard against using stale cached data. We study a variety of caching algorithms and a variety of methods for setting timestamps based on client mobility. We report results from simulation studies.