Robust Feature Set Research Articles

The Flexent™ Element Management System-Using web and object technologies to manage a wireless network

Telecommunications networks require a robust set of management features to ensure reliable service to end users. Element management features are needed to monitor, configure, and control the operation of each network element. The Flexent™ Element Management System (EMS) provides element-level management features to administer the collection of network elements constituting Lucent Technologies' Flexent/AUTOPLEX® wireless network, which provides wireless communications services to Lucent customers worldwide. Web and object-oriented technologies, commonly used within large enterprise networks, may also be used within element management applications for telecommunications networks. The Flexent EMS leverages standard management protocols, commercial object-based technology, and an intuitive Web-based user interface to administer the Flexent/AUTOPLEX wireless network. This paper describes the architecture and features of the Flexent EMS, discusses the unique challenges of element management for wireless networks, and shows how the Flexent EMS uses Internet Web technologies, object-oriented technologies, and usability engineering techniques to address these challenges.

Enhanced target identification using co-located acoustic and seismic sensors

Feature extraction and fusion of target signatures is of great importance in the detection, tracking, and identification of tracked and wheeled vehicles. Current research is centered in developing a set of robust features which will allow classification with high confidence regardless of various environmental conditions. The performance of boundary decision and nonparametric classifier topologies is also of special interest for the acoustic/seismic feature space. Acoustic and seismic data collected using co-located sensors have been processed using harmonic line association [J. A. Robertson, IIT Research Institute, in-house report] and higher order statistic [F. B. Shin and D. H. Kil, Proceedings of the International Conference on Image Processing, 3 (1995)] feature extraction techniques for classification. Performance results of the feature fusion will be presented as well as several classifier architectures. Real-time performance will also be presented [N. Srour, Remote Netted Acoustic Detection System, Army Res. Lab., May (1995)].

Feature analysis and neural network-based classification of speech under stress

It is well known that the variability in speech production due to task-induced stress contributes significantly to loss in speech processing algorithm performance. If an algorithm could be formulated that detects the presence of stress in speech, then such knowledge could be used to monitor speaker state, improve the naturalness of speech coding algorithms, or increase the robustness of speech recognizers. The goal in this study is to consider several speech features as potential stress-sensitive relayers using a previously established stressed speech database (SUSAS). The following speech parameters are considered: mel, delta-mel, delta-delta-mel, auto-correlation-mel, and cross-correlation-mel cepstral parameters. Next, an algorithm for speaker-dependent stress classification is formulated for the 11 stress conditions: angry, clear, cond50, cond70, fast, Lombard, loud, normal, question, slow, and soft. It is suggested that additional feature variations beyond neutral conditions reflect the perturbation of vocal tract articulator movement under stressed conditions. Given a robust set of features, a neural network-based classifier is formulated based on an extended delta-bar-delta learning rule. The performance is considered for the following three test scenarios: monopartition (nontargeted) and tripartition (both nontargeted and targeted) input feature vectors.