Abstract

Abstract

Abstract
This work presents an experimental study on a personal computer based implementation of a 64 kbit/s voice communication system over the Ethernet local network. Voice/data integration is achieved without resorting to any new packet switching protocol and without any changes to the CSMA/CD protocol specifications. System performance is evaluated according to its average transmission delay and percentage of packet loss as a function of network. load, packet size, buffers size, and traffic characteristics. An estimate of the maximum number of voice channels that can be supported without significant degradation of the reconstructed signal is also presented. A reference is made to synchronisation problems in packet voice systems and a method to eliminate clock drift is suggested. A non packet discarding reconstruction algorithm for a system with no silence detection is proposed. Finally, hardware dependent issues are also discussed.

Abstract
The relevance of scalp EEG recording on the selection of patients for epilepsy surgery usually is based on the concordance between the location of the epileptogenic focus and the presumed ictal origin of clinical seizures visually identified or video recorded. Bilateralisation and/or spreading of the epileptiform events are among the causes of the relative lack of agreement between scalp topography and origin of EEG potentials. Computer methods mainly if they are adaptive and use multistrategic approaches improve the accuracy on the detection of the epileptogenic focus, extracting relevant information on the location and spreading of epileptiform events.

Abstract

Abstract