An ongoing trend within the telecommunications industry has been toward providing end-to-end (E2E) IP based services. Accordingly, the associated management perspective has shifted from management of services within a single service provider domain towards a multi-domain paradigm supporting an E2E service view in which relationships among service providers, suppliers and customers and the related service level agreements (SLAs) take on major significance. In order to achieve multi-domain management, it is essential that a service management framework be put in place that supports planning, fulfilling, and assuring E2E services, including management of Service Requests, Service Assurance, and SLAs. The collecting and fusing of information from multiple sources, many external to the service provider's domain, to support the required E2E perspective has become an essential process.

Recently, local exchange carriers (LEC) in the USA have announced plans for large-scale deployments of fiber to the premises (FTTP) for residential customers and small and medium businesses; however, FTTP will only be successful if LEC develop and offer services that leapfrog the competition and grow the revenue stream. In addition, LEC will have to evolve every part of the network, not just the access. This paper shows that new services, not FTTP alone, will be the driver for significant evolutionary changes in the network, including optical core networks, service elements, and signaling and control.

Each vertex of an undirected graph possesses a piece of information that must be sent to every other vertex. They communicate by sending bounded size packets of messages from one vertex to another. The authors describe parallel algorithms, which accomplish the desired tasks for six prominent architectures. The algorithms are optimal, or nearly so, in every case.

Consider a network consisting of units connected by links in which some units could be faulty. Suppose each unit has a message which must be transmitted to all other (fault-free) units. We present an algorithm for doing this in a network operating in a fully distributed manner that requires at most 3n logn+O(n) message transmissions by fault-free units. Among other things, our result can be used to devise an algorithm for distributed system level diagnosis which is more efficient than the best currently known algorithm for this purpose.>

This paper considers the following problem: Given a positive integer t and graph H, construct a graph G from H by adding a minimum number $\Delta ( t,H )$ (respectively, $\Delta ' ( t,H )$) of edges and an appropriate number of vertices, such that after removing any t vertices (respectively, t edges) from G the remaining graph contains H as a subgraph. This problem was motivated by the design of fault-tolerant interconnection networks for multiprocessor systems. The authors estimate $\Delta ( t,H )$ and $\Delta ' ( t,H )$ for the cycle, path, complete binary tree, grid, torus, and hypercube on n vertices.

Consider a network in which each unit initially knows only its own identity and the identity of its immediate neighbors. Suppose each unit has a message intended for all other units. The authors give a distributed algorithm to accomplish this in point-to-point networks which is optimal in the number of transmissions it requires. They also show that this algorithm accomplishes this efficiently for broadcast (radio) networks, although the problem of finding a solution with the least number of transmissions, in broadcast networks, is shown to be NP-hard.>

Abstract Consider a network consisting of units and links that connect pairs of units. Suppose each of the units possesses a unique message that is to be received by all other units. This is called gossiping. A related problem is that of census taking in which a particular unit has to receive every other unit's message. We study gossiping and census taking in a network in which the units communicate by transmitting packets of a fixed size, so that only a bounded number of messages can be sent in a single transmission. We also discuss the complexity of gossiping when the messages are of different sizes.

One of the main methods for designing data link protocols in broadcast networks is based on the allocation of transmission rights to nodes in timeslots guaranteeing collision-free access to the channel. Extensions of certain results on vertex coloring are used to establish bounds on the number of timeslots required for a network to finish transmitting a backlog of data. Such networks are also observed when they are structured as rings and trees. Distributed algorithms for the allocation of timeslots in these networks are presented.>