Abstract: An access platform pools a plurality of different signal processing resources, herein represented by different types of speech-coding algorithms. The access platform is coupled to the Internet, a local-exchange-carrier (LEC), and other communications facilities such as, but not limited to, a plurality of long-distance facilities provided by any one of a number of long distance carriers, e.g., AT&T. For each call through the access platform, one of the plurality of signal processing resources is allocated as a function of signal type either through signal detection or out-of-band signaling. For example, the access platform first determines if the call is an audio call or a non-audio (or data) call by detecting the type of signal. If the call is an audio call, the access platform switches in echo canceling resources. On the other hand, if the call is a data call, the access platform determines if a speech-coding algorithm is being used and, if necessary, switches in a compatible speech-coding resource.

Abstract: A method and apparatus is described for improving the efficiency of any machine that uses an algorithm that maps to a higher dimensional space in which a given set of vectors is used in a test phase. In particular, reduced set vectors are used. These reduced set vectors are different from the vectors in the set and are determined pursuant to an optimization approach other than the eigenvalue computation used for homogeneous quadratic kernels. An illustrative embodiment is described in the context of a support vector machine (SVM).

Abstract: A kernel-based method and apparatus includes a preprocessor, which operates on an input data in such a way as to provide invariance under some symmetry transformation.

Abstract: In a receiver, a received sync-word is fed into two or more filters, where each filter further comprises a matching filter and a sidelobe reduction filter. Each matching filter is matched to a particular subset of k samples of the known sync-word, e.g., k=3, and generates an m-tap estimate, where m.gtoreq.2. The output signal of each matching filter is applied to a respective sidelobe filter, which reduces sidelobes resulting from the imperfections of matching the received sync-word in the matching filters. As a result, from each received sync-word, each filter forms a channel impulse response estimate of the communications channel at more than one symbol time. Each channel impulse response estimate comprises the amplitude as well as the phase of the channel impulse response.

Abstract: A timing technique for a Quantization-Level-Sampling (QLS) modem puts timing information in a downstream pulsed signal transmitted from the public switched telephone network (PSTN) to the QLS modem. In response to this timing information, the QLS modem synchronizes to the network sampling clock in the PSTN. In particular, the pulsed signal includes data-bearing samples, which were provided by a far-end QLS modem, and at least one non-user-data-bearing (NUDB) sample in which the level of this NUDB sample periodically alternates. The QLS modem extracts timing information from this periodic alternating signal level to synchronize the QLS modem to the network sampling clock.

Abstract: A blind convergence technique is restricted to using a subset of equalizer output samples. Illustratively, a receiver implements a windowed MMA approach. In this windowed MMA approach, a sample window overlays the two-dimensional plane representing the set of equalizer output samples. Only those equalizer output samples appearing within the sample window are used during filter adaptation.

Abstract: A packet voice system includes an ATM Adaptation Layer Type 2 (AAL-2) and Service Specific Convergence Sublayer (SSCS) System comprising a transmitter and a receiver. A portion of the 32 codepoints of the five bit RES (or UUI) field of the AAL-2 header are predefined to indicate an extended header. This a prior definition is stored in tables in both the transmitter and the receiver. The extended header itself comprises an additional octet appended to the AAL-2 header. A sequence number is normally carried in the RES field for the packet stream of voice calls that require sequence numbers. When the extended header is used, the sequence number is part of the extended header and thus messages are conveyed from the transmitter to the receiver without causing an interruption to sequence number assignment messages The transmitter dynamically uses the extended header and discontinues use of the extended header after a predefined duration of time or upon receiving an acknowledgment from the receiver.

Abstract: A decision feedback equalizer (DFE) comprises a feed-forward filter and a feedback filter. Blind training of the DFE is performed using a statistical-based tap updating algorithm for the feed-forward filter, and a symbol-based type of tap updating algorithm for the feedback filter.

Abstract: A user accesses the Internet through a hybrid data communications system architecture in which upstream service is provided via the local loop of the public switched telephone network (PSTN) and downstream service is provided via a cable television (CATV) network. The Cable Modem Data Termination system (CMTS) includes a routing table. Upon detection of a CATV service interruption, the CMTS re-routes IP traffic destined for a destination IP address (IP1) associated with the user to a different IP address, or endpoint, as a function of the routing table. In this instance, the different IP address is associated with equipment that provides downstream IP service to the user via the local loop of the PSTN.

Abstract: A wireless receiver includes a variable attenuator, a signal-to-noise ratio (SNR estimator), and a received signal strength indicator (RSSI) level detector. The RSSI level of the received signal is used as a primary criterion on setting the attenuator status. Once the attenuator is set based upon the RSSI readings, an SNR check is made to determine the possible presence of IM. Then, further adjustment on the attenuator status is made according to the SNR check.

Abstract: A demodulator of a CDMA base station receiver comprises a bank of 64 noncoherent Walsh correlators, each of which are identified by an index value and provide a correlator output signal to a soft decision metric generator. The soft decision metric generator provides a decision metric, D.sub.k, as a function of the L highest energy correlator output signals and their corresponding index values, where L>1.

Abstract: A distributed processing telephone system for providing "plug-and-play" capability. In particular, a key telephone system comprises a plurality of telephone sets. Each telephone set is coupled to a least one common communications channel, or telephone line, and includes at least one tunable RF modem. There is no key service unit (KSU). That is, the system is KSU-less. Resources of the telephone system are allocated using a peer-to-peer protocol. For example, as each telephone set is newly added to the system, the new telephone set adaptively determines its own allocation of resources, e.g., intercom numbers, etc. During operation, each telephone set requests the appropriate resources from its peers.

Abstract: A head-end dynamically allocates bandwidth of a communications channel as a function of the type of communications traffic. In particular, the head-end communicates to subscriber stations via a broadband cable network using an access protocol, which is modified to provide a variable number of mini-slots and a variable number of data slots in each frame. Each mini-slot is used to request assignment of a data slot(s) to subscriber stations for the communication of information and, also, as a vehicle to resolve contention between subscriber stations. The head-end dynamically adjusts the number of mini-slots over a period time as a function of the type of communications traffic, e.g., bursty and isochronous traffic sources. Any variation in the number of mini-slots concomitantly effects the number of data slots available to communicate information. For example, less mini-slots provides more data slots.

Abstract: A digital cellular/personal communications service (PCS) application incorporates a logical link connection (LLC) server. In this approach, there are two ATM connections in the path of an AAL-2 connection: one between a base station and the LLC Server and the other between the LLC Server and a vocoder group. All LLCs from, or to, a given base station use a common ATM connection irrespective of the vocoder used at the other end. Similarly, all LLCs from, or to, a given vocoder group use a common ATM connection irrespective of the destination base station at the other end. At the LLC Server, LLC packets from many base stations destined for the same vocoder group are extracted and bundled into the ATM connection between the LLC Server and the destination vocoder group. Similar treatment is given to the packets originating at vocoders and destined for base stations.

Abstract: A full duplex, synchronous data set (10) includes primary signal processing circuitry which generates a modulated transmit data signal in response to serial data from a terminal interface (17). The modulated data signal is transmitted over a primary channel of a transmit line (11). The primary signal processing circuitry also receives modulated data signals from a primary channel of a receive line (12) and recovers therefrom a serial bit stream for presentation to the interface. The operating parameters of the primary signal processing circuitry are specified by a primary controller (30) over a plurality of buses (PA, PC, PD). The primary controller includes a microprocessor (310) and associated peripherals (315, 320, 325, 330, 335). The data set also includes secondary signal processing circuitry (40) which transmits and receives diagnostic and control information over respective secondary channels of the transmit line and receive lines.