Search

We present a new approach for improving efficiency and performance of language translation process by a translator. The goal of this research is to develop efficient translation system for any language through enhancing existing approaches of language translation. In reality, it requires huge amount of time and resources to cover all the languages as all the languages possess numerous words having multiple meanings and also have sentences having multiple grammatical structures while expressing the same meaning. Therefore, it remains a great challenge to perform right semantic analysis. In this paper, we focus on efficient semantic analysis. We demonstrate efficiency of our approach through experimentation covering Bengali and English translation.

Recent rise in aged population and chronic diseases, is placing increasing pressure on health care expenditure. Ubiquitous health care is regarded as a potential driver in reducing such health expenditure. Advancement in wireless communication and sensor technologies permits real-time acquisition, transmission, and processing of critical medical information for ubiquitous health-care applications. Hence, medical body area networks (MBANs) emerge as a key technology to facilitate ubiquitous health-care services. However, energy restriction of micro-battery of a sensor device holds back the development of MBANs and also makes MBANs vulnerable to different malicious attacks. In this paper, we introduce a new attack entitled Power Attack exploiting power constraint of the sensor devices. Power attack forces a sensor to die off due to lack of power supply. Besides, we also propose countermeasure for the presented power attack. Finally, we exhibit efficacy of our proposed countermeasure using experimental evaluation.

People with visual impairment use Braille as a medium of textual representation having palpable dots. While the more privileged among us use OCR based text readers as preferential alternative, the lesser privileged ones are still bundling with conventional Braille tools. Our endeavor in this paper is to develop a low-cost and easy-to-use solution for less-privileged people. Here, we propose EyePen, a system that enables less-privileged visually-impaired people reading Braille characters, printed using conventional ink-jet and laser printers. The use of printed Braille characters using ink-jet printers offers a cost-effective alternative in place of the more expensively printed conventional palpable Braille dots. Besides, EyePen demands a very short learning period and offers a smooth learning experience ensuring ease-of-use while being in operation. We perform user evaluation over our developed system to confirm these findings. The user evaluation results show that the accuracy of reading printed Braille characters using EyePen converges to 100% within a very short period of time for all users. Finally, we present a potential writing aid that can facilitate low-cost and easy writing of Braille characters on normal paper exploiting our system.

Indoor localization or zonification in disaster affected settings is a challenging research problem. Existing studies encompass localization and tracking of first-responders or fire fighters using wireless sensor networks. In addition to that, fast evacuation, routing, and planning have also been proposed. However, the problem of locating survivors or victims is yet to be explored to the full potential. State-of-the-art literature often employ infrastructure dependent solutions, for example, WiFi localization using WiFi access points exploiting fingerprinting techniques, Pedestrian Dead Reckoning (PDR) starting from known locations, etc. Owing to unpredictable and dynamic nature of disaster affected environments, infrastructure dependent solutions are seldom useful. Therefore, in this study, we propose an ad hoc WiFi zonification technique (named as AWZone) that is independent of any infrastructural settings. AWZone attempts to perform localization through exploiting commodity smartphones as a beaconing device and successively searching and narrowing down the search space. We perform two testbed experiments. The results reveal that, for a single survivor or victim, AWZone can identify the search space and estimate a location with an approximate 1.5m localization error through eliminating incorrect zones from a set of possible results.

In recent times, migrating from touch-based input devices to touchless input devices has become one of the major research problems considering the health hazards exposed in public usages (e.g., information kiosk) of the touch-based devices. The problem still remains mostly open, since a solution having the properties of being low-cost, low power consuming, robust, and easy-to-use, is yet to be proposed in the literature. As a remedy to this situation, in this paper, we propose a visible light based touchless input device (named Sporshohin), retaining all the properties mentioned above, for ensuring ubiquitous usage. Sporshohin recognizes predefined gestures performed by the user, through an underlying mechanism that exploits well-known reflection model over the visible light spectrum. The underlying mechanism is intelligent enough to guarantee robustness against varying environmental lighting conditions and undefined noisy gestures. Real implementation, involving interaction with diversified users, confirms high accuracy and usability along with existence of all the above mentioned properties. We show the emulation of Sporshohin as an input interface for an information kiosk. The attributes of Sporshohin exhibit its prospect of being used in diversified applications going beyond information kiosks.

Under-reporting is one of the main causes of failure to solve social problems, which obstruct national development in third-world countries. A one-stop online reporting system can facilitate minimizing the extent of under-reporting, which is yet to be developed for general people of third-world countries. Therefore, in this paper, we propose a generic online reporting system where one can submit report anonymously, even without registration. Our system aims at propagating the reports to the respective authorities such as law enforcement agencies, anti-corruption commission, city corporation, policy makers, human rights commissions, etc., after a reviewing process. The system will also publish the reports without disclosing identities of the reporters to disseminate the information among public and to collect the public opinions about the reports.

Air pollution is a serious threat to public health. The city residents often spend significant time in public transports owing to long traffic tailbacks. So, the air quality inside public transports is an important factor that needs to be considered for the welfare of city residents. In this paper, we study the air quality inside public transports in a metropolitan city from the perspective of a developing country. To capture the diversity of city life, we consider the variation in traffic pattern and public transports, and analyze the data in temporal and spatial domain. Our custom-built sensing module collects data at regular intervals and detects the presence of pollutants above recommendation level. Our temporal analysis shows that concentration of air pollutants inside different public transports do not remain consistent in peak and off-peak hours. Our analysis also reveals that public transports with lower fares are more exposed to air pollutants than those with higher fares.