Sharif Digital Repository

This paper proposes four methods for improving the performance of keyword spotting (KWS) systems. Keyword models are usually created by concatenating the phoneme HMMs and garbage models consist of all phonemes HMMs. We present the results of investigations involving the use of skips in states of keyword HMMs and we focus on improving the hit ratio; then for false alarm reduction in KWS we model the words that are similar to keywords and we create HMMs for highly frequent words. These models help to improve the performance of the filler model. Two post-processing steps based on phoneme and word probabilities are used on the results of KWS to reduce the false alarms. We evaluate the... 

Presently, the urgency of creating efficient system to tackle natural hazards such as tsunamis, earthquakes, landslides and floods to cite just a few, have originated that Wireless Sensor Networks (WSNs) play an increasingly important role in the natural hazards scenario as an efficient early warning system, sometimes in synergy with satellites-GMES (Global Monitoring Earth Systems) and also with mobile robotics. Their integration into the web scenario, giving rise to the sensor-web technology, provides even more power to the intrinsic strengths of WSNs. The WSNs application in post-disaster scenarios is also an emerging challenging issue. In this paper the fundamental technology concepts of... 

A significant challenge in spectrum sensing is to lessen the signal to noise ratio needed to detect the presence of primary users while the noise level may also be unknown. To meet this challenge, multi-antenna based techniques possess a greater efficiency compared to other algorithms. In a typical compact multi-antenna system, due to small interelement spacing, mutual coupling between thermal noises of adjacent receivers is significant. In this paper, unlike most of the spectrum sensing algorithms which assume spatially uncorrelated noise, the noises on the adjacent antennas can have arbitrary correlations. Also, in contrast to some other algorithms, no prior assumption is made on the... 

Today, from information security perspective, prevention methods are not enough solely. Early Warning Systems (EWSs) are in the category of reactive methods. These systems are complementing Intrusion Detection Systems (IDSs) where their main goals include early detection of potential malicious behavior in large scale environments such as national level. An important process in EWSs is the analysis and correlation of alerts aggregated from the installed sensors (e.g., IDSs, IP telescopes, and botnet detection systems). In this paper, an efficient framework for alert correlation in EWSs is proposed. The framework includes a correlation scheme based on a combination of statistical and stream... 

Vacuum tubes are widely used for various applications. These vacuum tubes are supplied by high voltage power supplies. The amount of delivered energy from the high voltage power supply to the vacuum tube is an important issue during the vacuum arc in the tube. The protection mechanism consists of a shunt crowbar which diverts the fault current from the tube to itself as a parallel path. Detection of the vacuum arc is crucial and only one sensor is usually employed to detect the vacuum arc. This characteristic intensifies the interference susceptibility of the vacuum arc diagnosis system in a noisy environment. As a result of the noise, the arc detection system can report false alarms. False... 

In this paper a method for detecting out of vocabulary words in isolated word recognizers is introduced, our method utilized new kinds of confidence measure. After recognition task was completed and consequently confidence measure was extracted, a classifier would accept or reject result of recognition task using this CM. We used two different kinds of confidence measure where for extracting each one a different information source was used. Amount of competition between hypotheses through the recognition task was used for extracting first CM. The second one was extracted using information about manner of distribution of feature vectors in the states of winner HMM model. Both of these CMs... 

Multiple target detection in conventional PCL (Passive Coherent Location) radar systems by means of Ambiguity Function Processing and without any powerful algorithm is impossible. Presented Gradual Clean Algorithm (GCA) is a new and simple method, which could provide the perfect detection in noisy environment and heavy clutter condition up to SNR=-15 dB and SCR=-100 dB (clutter is assumed as a distributed target with large RCS). Determining the number of real targets and keeping the false alarm probability approximately constant are the presented algorithm capabilities