Abstract: A given network of nodes that are interconnected by links having corresponding capacities has each link's capacity divided into working capacity and restoration capacity without a priori information about network traffic characteristics. Allocation of working capacity and restoration capacity for the network might be optimized by characterization of the network in accordance with a linear programming problem (LPP) subject to network constraints and then generating a solution to the LPP either exactly or with an approximation. Partitioning the capacity of each link in the network into working and restoration capacities minimizes the restoration capacity overhead in the network to allow for higher network utilization.

Abstract: A computer-implemented method of computing throughput of a data-routing scheme for a network of nodes interconnected by links and having at least one ingress point and at least one egress point. The method includes: deriving a polynomial-size linear program from a combination of a first linear program and a second linear program and solving the polynomial-size linear program. The first linear program has infinite constraints and minimizes maximum-link utilization of a link in a path between the ingress point and the egress point. The second linear program determines whether any constraint of the first linear program is violated.

Abstract: A network of nodes interconnected by links, such as a wavelength division multiplexing (WDM) ring network, is modeled by a delayed graph in which propagation delay between nodes is accounted for. Given the delayed graph, a traffic matrix for an admissible traffic pattern is scheduled over a frame period. Scheduling of the traffic matrix formulates a set of scheduling constraints based on the delays and decomposes the traffic matrix into a set of transmission matrices. Each of the set of transmission matrices is scheduled over the frame period.

Abstract: A method and system for generating a bloom filter by mapping into respective groups each of a plurality of initial keys according to a first hash function and mapping each group hashed key into a bloom filter using k respective hash functions.

Abstract: A method of estimating per-flow traffic in a network of nodes interconnected by links splits the incoming traffic stream to a node into a number of parallel sub-streams by flow assignment using a hash function. Since each flow in the incoming traffic stream constitutes a larger fraction of the traffic in its sub-stream, the probability of obtaining two-runs increases. The number of two-runs per flow is then employed to generate an estimate of the traffic of the flow.

Abstract: A lightweight probabilistic mechanism used to estimate the number of active flows, which estimate is used to determine the probability of admitting a new flow into the network. In one embodiment, a method for controlling admission of new flows at a node in a network of nodes interconnected by links includes: (a) for each of a plurality of incoming packets arriving at the node, each incoming packet corresponding to an active flow traversing the node: (a1) randomly selecting a packet from an output buffer of the node; (a2) determining whether the incoming packet is from the same active flow as the randomly-selected packet; and (a3) updating an estimate of the number of active flows traversing the node based on the determination of step (a2); and (b) determining whether to admit or drop part or all of a new flow at the node based on the estimated number of active flows traversing the node.

Abstract: A fast and efficient method for network re-optimization is described. In one embodiment, the technique uses an algorithm that requires limited aggregate information to continually monitor the network for re-optimization opportunities. The re-optimization of the present invention enables a network to accommodate more traffic without adding capacity. The re-optimization of the present invention is performed only when network efficiency benefits above a predetermined threshold may be obtained. In addition, a least-cost set of sessions is determined to perform the re-routing of the present invention to attain the desired network efficiency.

Abstract: The invention includes an apparatus and method for switching packets through a switching fabric. The apparatus includes a plurality of input ports and output ports for receiving arriving packets and transmitting departing packets, a switching fabric for switching packets from the input ports to the output ports, and a plurality of schedulers controlling switching of packets through the switching fabric. The switching fabric includes a plurality of virtual output queues associated with a respective plurality of input-output port pairs. One of the schedulers is active during each of a plurality of timeslots. The one of the schedulers active during a current timeslot provides a packet schedule to the switching fabric for switching packets through the switching fabric during the current timeslot.

Abstract: We formulate the network-wide traffic measurement/analysis problem as a series of set-cardinality-determination (SCD) problems. By leveraging recent advances in probabilistic distinct sample counting techniques, the set-cardinalities, and thus, the network-wide traffic measurements of interest can be computed in a distributed manner via the exchange of extremely light-weight traffic digests (TD's) amongst the network nodes, i.e. the routers. A TD for N packets only requires O(loglog N) bits of memory storage. The computation of such O(loglog N)-sized TD is also amenable for efficient hardware implementation at wire-speed of 10 Gbps and beyond. Given the small size of the TD's, it is possible to distribute nodal TD's to all routers within a domain by piggybacking them as opaque data objects inside existing control messages, such as OSPF link-state packets (LSPs) or I-BGP control messages.

Abstract: A network-wide traffic measurement/analysis problem is formulated as a series of set-cardinality-determination (SCD) problems, using probabilistic distinct sample counting techniques to compute network-wide traffic measurements of interest in a distributed manner via the exchange of light-weight traffic digests (TD's) amongst network nodes/routers. A TD for N packets uses only requires O(loglog N) bits of memory storage, making it possible to distribute nodal TD's to all routers within a domain by piggybacking them as opaque data objects inside existing control messages, such as OSPF link-state packets (LSPs) or I-BGP control messages. A router receiving the TD's can estimate the traffic measurements of interest for each of its local links by solving a series of set-cardinality-determination problems.

Abstract: The invention includes a method and apparatus for determining a routing table for use in a network comprising a plurality of type-one nodes and a plurality of type-two nodes where the type-two nodes use respective type-two routing tables having one next-hop node associated with each of a plurality of destination nodes of the network. The method includes obtaining, from the type-one nodes and the type-two nodes, link status information associated with each of a plurality of links of the network, obtaining traffic measurement information from the type-one nodes and the type-two nodes, and determining, using the link status and traffic measurement information, type-one routing tables for the respective type-one nodes. The type-one routing tables are determined as a group in a substantially coordinated manner, and each type one routing table includes a plurality of next-hop nodes associated with each of the destination nodes.

Abstract: A packet network employs restorable routing with service level guarantees. Restorable routing generates two disjoint paths through a network of nodes interconnected by links for a connection request demand between and ingress-egress node pair. Restorable routing employs minimum interference criteria to generate the two disjoint paths such that two disjoint paths cause little or no interference with demands of future connection requests between different ingress-egress pairs. Restorable routing generates maximum 2-route flows for the network ingress-egress node pairs to determine corresponding sets of 2-critical links. A reduced network is formed, its links are weighted based on criticality indices generated from the sets of 2-critical links, and the relatively optimal two disjoint paths are computed for the connection request. One of the two disjoint paths is selected as an active path for routing data of the connection request, and the other disjoint path is selected as the backup path.

Abstract: A network of nodes interconnected by links has content filtering specified at certain nodes, and routing of packet connections through the network is generated based on the specified content-filtering nodes. The network is specified via a content-filtering node placement method and a network-capacity maximization method so as to apply content filtering to packets for substantially all traffic (packet streams) carried by the network.

Abstract: A switch schedules guaranteed-bandwidth, low-jitter-traffic characterized by a guaranteed rate table (GRT) method. A rate matrix generated from collected provisioning information is decomposed into schedule tables by a low jitter (LJ) decomposition method. The LJ decomposition method imposes a set of constraints for the schedule tables: schedule tables are partial permutation matrices, weighted sum of the partial permutation matrices is greater than or equal to the weighted sum of the rate matrix, and each entry in the rate matrix belongs to one element of the LJ decomposition schedule matrices. An integer LJ decomposition programming problem is employed to generate the schedule tables that are scheduled for each time slot of the period of the switch. Schedule tables are selected in turn based upon selecting eligible tables having the earliest finishing time. If necessary, the rate matrix is updated prior to decomposition for a subsequent period.

Abstract: A scheme for guaranteeing network tunnel path (NTP). e.g., Internet Protocol (IP), service levels in optical (e.g., wavelength-division multiplex (WDM)) networks. The invention, in one embodiment, accounts for both network topology (e.g. switches, cross-connects, and links between nodes) and resource usage (e.g., available/provisioned link bandwidth and available/used wavelength paths), which information may be available from the IP and optical WDM protocol layers, in particular (i) router capacities and (ii) presence or absence of wavelength conversion capability of optical cross-connect at each node. A determination is made whether to route an arriving request for an NTP over existing topology by computing “good” routes in accordance with a defined metric, or to open a new, available optical wavelength path by provisioning nodes of the topology for, and computing “good” routes for new wavelength paths. As many requests as possible are identified without a priori information of future requests.

Abstract: Capacity design of an optical network for demands of fast path restorable (FPR) connections forms a linear programming sizing problem for a optimal routing. A dual of the linear programming sizing problem is formed and solved with an approximation algorithm. Edge lengths are initialized based on i) the inverse of the edge's capacity and ii) a scalar constant. Then, the approximation algorithm proceeds in phases to route each commodity over the edges of a graph. During each phase, the demand's flow is sent from the source to destination via multiple iterations. During each iteration, the set of shortest disjoint paths from the source to the destination is determined, a portion of the flow is sent, and the lengths of the edges that carry the flow are updated. The value employed to scale the network is generated after the last phase from the maximum ratio of edge flow to edge capacity.

Abstract: Restorable paths in an information network are established in response to arriving traffic requests. Requests are received at a first node of the network for transmission of traffic to a second node of the network, and each request specifies a desired transmission bandwidth for an active and a backup path to be established between the nodes. Potential active links for an active path are identified in response to a given request, and potential backup links to form a backup path for restoring the active path, are also identified in response to the given request. An active and a backup path are then formulated for each given request from among the potential active links and the potential backup links that were identified in response to the given request.

Abstract: A packet network of interconnected nodes employing dynamic backup routing of a Network Tunnel Path (NTP) allocates an active and backup path to the NTP based upon detection of a network failure. Dynamic backup routing employs local restoration to determine the allocation of, and, in operation, to switch between, a primary/active path and a secondary/backup path. Switching from the active path is based on a backup path determined with iterative shortest-path computations with link weights assigned based on the cost of using a link to backup a given link. Costs may be assigned based on single-link failure or single element (node or link) failure. Link weights are derived by assigning usage costs to links for inclusion in a backup path, and minimizing the costs with respect to a predefined criterion.

Abstract: A network element maintains failure information for a packet-based network and usage information for a backup path. Upon receipt of a new demand, with an associated bandwidth, d, the network element determines if the backup path can be shared as a function of the failure information and the usage information associated with the backup path.

Abstract: Capacity design of an optical network for demands of fast path restorable (FPR) connections forms a linear programming sizing problem for a optimal routing. A dual of the linear programming sizing problem is formed and solved with an approximation algorithm. Edge lengths are initialized based on i) the inverse of the edge's capacity and ii) a scalar constant. Then, the approximation algorithm proceeds in phases to route each commodity over the edges of a graph. During each phase, the demand's flow is sent from the source to destination via multiple iterations. During each iteration, the set of shortest disjoint paths from the source to the destination is determined, a portion of the flow is sent, and the lengths of the edges that carry the flow are updated. The value employed to scale the network is generated after the last phase from the maximum ratio of edge flow to edge capacity.