Fast packet switching

About: Fast packet switching is a research topic. Over the lifetime, 5641 publications have been published within this topic receiving 111603 citations.

Crossbar switching matrix with broadcast buffering

Abstract: A novel crossbar switching matrix that improves the transmission of variable length broadcast packets by greatly reducing transmission latency time. Unicast traffic is halted in the middle of packet transmission operations and the broadcast packet is transmitted. Once the broadcast packet has finished transmission, unicast packet transmission resumes without any loss of data. A unicast buffer is used to store the unicast packet while the broadcast packet is being transmitted. A broadcast buffer is used to buffer the broadcast packet as it egresses from the switch matrix. In this fashion, the broadcast information is given high priority and passes quickly through the switch without the large delays associated with the prior art switch matrixes. The crossbar switch matrix of the present invention is applicable to any switching matrix adapted to switch variable length data units.

A token based protocol for integrated packet and circuit switching in WDM rings

Abstract: The paper proposes a networking solution that represents the first step towards the integration of circuit and packet switching techniques directly in the WDM layer, considered to be a fundamental milestone on the way of creating the next generation Internet (NGI). Focusing on ring networks using extant optical devices, the proposed solution allows for the all-optical transmission of individual packets and circuits via a "tell and go" establishment of lightpaths (wavelength routes). The integration of the two alternative switching techniques is obtained by a novel multi-token control protocol shown to maximize bandwidth utilization, minimize latency and scale ideally in the network.

Robust speech recognition in burst-like packet loss

Abstract: This paper examines problems associated with performing speech recognition over mobile and IP networks. The main problems are identified as codec-based distortion and from speech vectors being lost from packet loss in the network. A realistic model for packet loss is developed, based on a three state Markov model and is shown to be capable of simulating the burst-like nature of packet loss. A two stage packet loss detection and estimation scheme is proposed and is shown to improve the recognition performance in the event of feature vectors being lost. Results from the Aurora database show that burst-like packet loss reduces the digit accuracy from 99% to 57% at 50% packet loss. Estimation of the lost packets recovers the performance to 77%.

Technique for implementing an admission control scheme for data flows

Abstract: A technique for implementing an admission control scheme for data flows is disclosed. In one embodiment, the technique is realized by determining a current error value in response to a queue size in a packet buffer; generating a packet drop probability based at least in part on the current error value; receiving a data packet having a data type; and determining whether to reject or accept the received data packet at a queue based at least in part on the packet drop probability and a predetermined flow rejection threshold. In addition, the technique may involve determining whether a randomly generated number is less than or equal to the packet drop probability and determining whether a count variable is greater than or equal to an inter-drop interval.

A neural-network contention controller for packet switching networks

Abstract: A novel approach to solving the output contention in packet switching networks with synchronous switching mode is presented. A contention controller has been designed based on the K-winner-take-all neural-network technique with a speedup factor to achieve a real-time computation of a nonblocking switching high-speed high-capacity packet switch without packet loss. Simulation results for evaluation of the performance of the K-winner network controller with 10 neurons are presented to study the constraints of the "frozen state" as well as those of same initial state. An optoelectronic contention controller constructed from a K-winner neural network is proposed. >