Abstract: A method segments a video. Audio frames of the video are classified with labels. Dominant labels are assigned to successive time intervals of consecutive labels. A semantic description is constructed for sliding time windows of the successive time intervals, in which the sliding time windows overlap in time, and the semantic description for each time window is a transition matrix determined from the dominant labels of the time intervals. A marker is determined from the transition matrices, in which a frequency of occurrence of the marker is between a low frequency threshold and a high frequency threshold. Then, the video is segmented at the locations of the markers.

Abstract: A computer implemented method detects scene boundaries in videos by first extracting feature vectors from videos of different genres. The feature vectors are then classified as scene boundaries using a support vector machine. The support vector machine is trained to be independent of the different genres of the videos.

Abstract: During operation a node (500) that has associated with a network (100) will periodically broadcast a beacon (303) during a beacon interval as part of a superframe. The beacon will be periodically broadcast for a first period of time, based on a time it takes for nodes to associate with the network. After the first period of time has passed, the beacon will be switched off for a second period of time, after which, the beacon will again be periodically broadcast.

Abstract: A location technique is utilized where channel-model parameters are originally estimated prior to location taking place. Location then takes place using a first set of known-located nodes, and the channel-model parameters are updated based on the distances resulting from the location estimate. Once the channel-model parameters have been updated, location again takes place using a second set of known-located nodes, node distances are calculated based on the produced locations and the channel-model parameters are again updated. This process continues until no significant change is observed between the previous and the newly estimated location, or until a maximum number of iterations is reached.

Abstract: A method and apparatus for determining the path loss model of an object within a wireless communication system is provided herein. During operation, a path loss model for a node is generated based on path loss values and corresponding numbers of neighbors of the said node. The path loss model is used to determine a relationship between path loss and distance. With this relationship established, distances to known-located nodes may be obtained by obtaining a path loss to the known-located node. From these distances, a node can then be located.

Abstract: A method and apparatus for determining the relative locations of nodes in a wireless network is disclosed. A list of neighboring nodes of a master node of the wireless network is generated and, from this list, a list of independent nodes is generated. An adjacency table for the independent nodes is then generated by examining the neighbors of dependent nodes. Two independent nodes are determined to be adjacent nodes only if they are the only neighboring nodes of a dependent node. Relative locations of nodes in wireless network may be determined from the adjacency table.

Abstract: A computer implemented method detects scene boundaries in videos by first extracting feature vectors from videos of different genres. The feature vectors are then classified as scene boundaries using a support vector machine. The support vector machine is trained to be independent of the different genres of the videos.

Abstract: A method segments a video. Audio frames of the video are classified with labels. Dominant labels are assigned to successive time intervals of consecutive labels. A semantic description is constructed for sliding time windows of the successive time intervals, in which the sliding time windows overlap in time, and the semantic description for each time window is a transition matrix determined from the dominant labels of the time intervals. A marker is determined from the transition matrices, in which a frequency of occurrence of the marker is between a low frequency threshold and a high frequency threshold. Then, the video is segmented at the locations of the markers.

Abstract: A method for peer-to-peer ranging and discovery of a rigid body existing in a scatternet having piconets and nodes includes the steps of defining a node (12) in a piconet (10) to be a piconet controller (PNC) having controller functions, locating a rigid body seed including the node (12), and discovering a rigid body by sequentially downloading controller functions of the piconet controller (12) to at least one border node. Also provided is a communications node including a receiver for receiving communications from other communications nodes in a communications range (R), a transmitter for sending communications to other communications nodes in the communications range (R), a memory storing at least ranging information and a unique identification for describing the node, and a processor connected to the receiver, to the transmitter, and to the memory, the processor being programmed to carry out the method according to the present invention.

Abstract: A location technique is utilized within a wireless communication system where power measurements from various nodes are weighted to determine a node's location. An iterative location process is executed where as more and more nodes are heard from, a node will re-locate itself in order to improve location accuracy.

Abstract: A method for locating nodes in a multi-hop sensor network forms a rigid body (RB1, RB2, RB3) and, from the nodes, utilizes the rigid body to decide if a node is locatable. The method obtains a reduced order model (ROM) of the network by categorizing all of the nodes by location status, grouping them based upon the categorizations, and defining and identifying a rigid body from a group. The method further simplifies determinability of node location by forming the rigid body from the nodes based upon the categorized location status. To locate the nodes, the nodes are separated from one another into subsets dependent upon characteristics (100). Then, groups are formed from one subset (200) and the rigid body is formed from a group (300). The ROM is formed from the rigid body (400) and a location capability of the rigid body is evaluated based upon the ROM (500).

Abstract: A method and system is provided for determining a location for each of a plurality of units, which are sub-divided into more than one sub-net groupings, that each include two or more units. Within each sub-net grouping, measured range information between units within the sub-nets and one or more reference units is gathered in at least a selected one of the units. The selected one of the units then estimates a location for each of the units, which minimizes any error in the measured ranges between the units in each of the corresponding sub-nets.

Abstract: A method for reducing communications in a peer-to-peer wireless network having nodes includes sending an RTS-TOA ranging communication from a first node (N0) in a node group to another node (Nr, I) in the group, receiving the RTS-TOA communication with the second node (Nr, I) and sending a multi-cast CTS-TOA/RTS-TOA ranging communication from the second node (Nr, I) as a reply to the received RTS-TOA communication, and successively and sequentially repeating the multi-cast sending step for each of the nodes. A CTS-TOA message is a reply to the received RTS-TOA message and, simultaneously, is an RTS-TOA ranging communication to a new destination node (J). The method applies for a randomized communications approach when node identifications are not sequenced and to a sequential approach where nodes are sequenced, and also applies to completely connected and multi-hop networks. Also provided is a node for carrying out the method of the present invention.

Abstract: During operation a node (500) that has associated with a network (100) will periodically broadcast a beacon (303) during a beacon interval as part of a superframe. The beacon will be periodically broadcast for a first period of time, based on a time it takes for nodes to associate with the network. After the first period of time has passed, the beacon will be switched off for a second period of time, after which, the beacon will again be periodically broadcast.

Abstract: A method and location determination module is provided for determining a location of one of a plurality of units using neighbor lists. Each unit is communicatively coupled to at least some of the other plurality of units, where at least some of the plurality of units are reference units, whose locations are known. The units communicate with other nearby units within communication range, to establish neighbor lists. A unit to be located then identifies an aggregate value corresponding to the number of occurrences of the reference units in the neighbor list of the unit to be located and the neighbor lists of each of a group of associated units. The location of the unit to be located is then determined, based upon the known locations of the reference units and the number of identified occurrences of the reference units in the corresponding neighbor lists.

Abstract: A method and apparatus for determining the location of a node within a communication system is provided herein. During operation, reference nodes (105) having known locations are utilized to locate “blind” nodes (104) whose location is to be determined. More particularly, a blind node (104) wishing to determine its location will measure a plurality of path losses between itself and a plurality of reference nodes (105). These reference nodes' locations will then be mathematically weighted by the path loss between these reference nodes (105) and the blind node (104). The location of the blind node (104) is a sum of the mathematically weighted reference nodes' locations.

Abstract: A method for locating nodes in a multi-hop sensor network forms a rigid body (RB1, RB2, RB3) and, from the nodes, utilizes the rigid body to decide if a node is locatable. The method obtains a reduced order model (ROM) of the network by categorizing all of the nodes by location status, grouping them based upon the categorizations, and defining and identifying a rigid body from a group. To locate the nodes, the nodes are separated from one another into subsets dependent upon characteristics (1100). Then, super groups are formed from one subset (1200) and the sub-groups are formed from each super group (1300). The rigid body core is formed (1400) and the resultant rigid body is determined (1500). The ROM is formed from the rigid body (1600) and a location capability of the rigid body is evaluated based upon the ROM (1700).

Abstract: A location technique is utilized within a wireless communication system where power measurements from various nodes are weighted to determine a node's location. An iterative location process is executed where as more and more nodes are heard from, a node will re-locate itself in order to improve location accuracy.

Abstract: A method and apparatus for determining the path loss model of an object within a wireless communication system is provided herein. During operation, a path loss model for a node is generated based on path loss values and corresponding numbers of neighbors of the said node. The path loss model is used to determine a relationship between path loss and distance. With this relationship established, distances to known-located nodes may be obtained by obtaining a path loss to the known-located node. From these distances, a node can then be located.

Abstract: A location technique is utilized where channel-model parameters are originally estimated prior to location taking place. Location then takes place using a first set of known-located nodes, and the channel-model parameters are updated based on the distances resulting from the location estimate. Once the channel-model parameters have been updated, location again takes place using a second set of known-located nodes, node distances are calculated based on the produced locations and the channel-model parameters are again updated. This process continues until no significant change is observed between the previous and the newly estimated location, or until a maximum number of iterations is reached.