Viswanath Subramanian Ramesh ¹, P. Saranya ¹, Ruchira Kapoor ¹

Affiliations
1 SRM University, Kattankulathur, Kanchipuram - 603 203, Tamil Nadu, IN

Abstract

Objectives: To develop an application in order to make the examination process easier for the visually impaired and also to reduce the conventional examination setup overheads. Methods: The conversion of text to speech and vice versa has been carried out by existing modules in the android platform. Adaptive examination is carried out based on performance analysis of the candidate by assessing the cognitive ability (reaction time) which is implemented through the concept of time stamping. Findings: The equation formulated for making the examination adaptive, returned authentic and error proof values which could be used for setting the difficulty of the consequent sections of the question paper, by comparison with predetermined ideal performance value. Also it was evident that the setup process was relatively simpler than a pen-paper examination setup. The speech rate of the audio module had to be personalized according to the candidate’s requirement. By incorporating the changes and handling the runtime glitches, it was found that application made the examination process hassle-free for visually impaired students. Improvements/Applications: Objective analysis of cognitive ability has been made possible through the newly formulated equation. The result obtained can be applied in multiple domains.

Keywords

Android, Automatic Speech Recognition (ASR), Adaptive Examination, HMM, Speech Recognition, Time Stamping.

Full Text

   

Bhavya Sharma ¹, Adit A. Pillai ¹, P. Saranya ¹

Affiliations
1 Computer Science and Engineering department, SRM University, Kattankulathur, Kanchipuram - 603 203, Tamil Nadu, IN

Abstract

Objectives: The primary objective of the paper is to propose an optimized data extraction technique to obtain data from the Wireless Sensor Networks (WSNs)using ‘Sensor Head Algorithm’ and Data Fusion. Methods/Statistical Analysis: The proposed idea works in a three-layered architecture which includes the deployed sensors, the selected Sensor Head and the Static Data Sink. The Sensor Head Algorithm is used to select a sensor head that collects data from all sensors. The Data Fusion algorithm results in generation of unified data which is used to derive the output. Finally, the fused data is transmitted from the Sensor Head to the Static Data sink. Findings: A sample data was used to verify the working of the model using simulations. On the application of Sensor Head Algorithm, the sensors are assigned priorities on the basis of their residual energies. The sensor with highest residual energy becomes the Sensor Head. All the sensors transmit their data to the Sensor Head where the Data Fusion takes place later. Data Fusion initially removes redundancy in from the collected data making the processing simpler and saving the energy. The resultant data is then used to generate fused inference using Fuzzy Logic Controller (FLC). These inferences are recorded and later used to generate final output. The proposed idea ensures reduction of load on the system, simplifies the previously proposed methods, reduces latency in data collection and improves network lifetime. This data, finally transmitted to the static data sink, can be used for finding inferences or reaching a conclusion about the target area in the desired way. Application/Improvements: In the current paper, we have proposed an idea which we shall use to develop a fully functional model. Since, the proposed technique is concerned with data extraction, it can be applied to every situation where the Wireless Sensor Network is deployed.

Keywords

Data Fusion, Fuzzy Logic Controller, Residual Energy, Sensor Head Algorithm, Static Data Sink, Wireless Sensor Networks (WSNs).

Full Text