1.  Analysis of charge-shared Matchline sensing schemes and current race scheme in high-speed ternary content addressable memory (TCAM)

      Authors: Raqqib Bin Kadir; Ahmed Selim Anwar; Muntasim-Ul-Haque 

     Published in: 2016 International Conference on Innovations in Science, Engineering and Technology (ICISET)

Abstract:

Due to emerging of hi-speed communication larger TCAMs with higher speed is needed. A comparative analysis of different Matchline sensing schemes in high speed Ternary content addressable memory (TCAM) is presented in this paper. With the conventional current race scheme, two different methods of charge sharing matchline sensing schemes are being analyzed. The power is distributed along the matchline by dividing it and sharing charge. This two charge shared schemes also improve search time and voltage margin. Simulations are performed using 180nm 1.8V CMOS logic in HSPICE.

2.  Challenges in Implementing Machine Learning- Driven IoT Solutions in Semiconductor Design and Wireless Communication System 

      Authors: Ahmed Selim Anwar; Sujan Kumar Bhowmik; Raqqib Bin Kadir; Muntasim-Ul-Haque 

     Published in: International Journal on Recent and Innovation Trends in Computing and Communication(IJRITCC)

Abstract:

The integration of machine learning and Internet of Things has great potential to revolutionize several industries including semiconductor design and wireless communication techniques as it can. However, the application of ML in the IoT poses great problems such as compatibility with existing systems, real-time decision making, scalability and cost. This research seeks to investigate these challenges by identifying major technical and operational impediments that hinder IoT based ML application in these industries.

The current research adopted the quantitative research approach and the data was collected through an online questionnaire  completed by the professionals in the semiconductor and wireless communication industries. Questionnaire was distributed and filled by 300 participants, to evaluate their experience about ML-IoT integration. The survey encompasses a number of aspects such as technical issues, practical problems, the questions of expansion of the usage and questions of data protection. Quantitative data were analyzed descriptively and inferentially employing Chi-square test, ANOVA, multiple regression analysis, correlation analysis.These techniques enabled the assessment of the issues and interactions between the factors of industry focus, company size and ML IoT implementation.

The article concluded that the main challenges still persists and they are chiefly evidenced in the issues of legacy system integration whereby semiconductor design still finds itself in the lager of hauling old architectures that cannot support computation of today’s convolutional ML algorithms. The identified key issue in wireless communication environment was real-time decision-making because it could not afford a time delay in processing of large amount of data when required. The issue of scalability was identified to be widely affecting both industries as they attempt to handle the increasing amount of IoT data in efficient and performant manners. Further, the data privacy and security were reported to be slightly higher in the wireless communication system and the participants called for enhanced legal safeguards and privacy-preserving methodologies in the ML domain. The regression analysis revealed that the difficulty level of the ML-IoT challenges was dependent on the size of the organization and the number of years spent on such projects: While larger organizations had more potential to come to terms with the issues of scalability of the projects, they required in order to advance, they still struggled with the costs involved in the projects and the matter of data privacy. The article suggests that the use of techniques like edge computing and federated learning can eliminate the challenges posed by real-time processing while cloud environments can also be the possibility of cost-saving. More emphasis must be placed on security solutions that facilitate data privacy such as privacy-preserving technologies and commitment to more technological advances for creating more scalable ML models to be used in IoT systems.

3.  Performance Comparison of Charge-Shared Matchline Sensing Schemes in High-Speed Ternary Content Addressable Memory (TCAM)

      Authors: Raqqib Bin Kadir; Ahmed Selim Anwar; Muntasim-Ul-Haque 

Abstract:

Ternary content addressable memory (TCAM) is a memory that offers high speed table look-up capability for applications such as internet protocol (IP) packet forwarding and classification in network routers. A performance comparison between different Matchline sensing schemes in high speed Ternary content addressable memory (TCAM) is presented in this thesis. With the conventional current race scheme two different charge shared schemes are being compared. By segmentation of Matchline and then charge sharing reduces the power to some extent. This two charge shared schemes also improves search time and voltage margin. Simulations are performed using 180nm 1.8V CMOS logic in HSPICE. By changing the properties of different transistors, the simulated outputs were compared for better performance measurements. 

4. Comparison of Heat Transfer Through a Circular Pipe Using Different Geometry Inside the Pipe 

Authors : Sheikh Fuzael Rahman, Md Irfan Khan, Amit Das, Raqqib Bin Kadir 

     Published in: International Journal of Scientific & Engineering Research

Abstract:

The purpose of this paper is to observe different rate of heat transfer while changing the surface area at different condition inside the circular pipe section. Insulation was given to compare the heat transfer rate with non-insulated ones. The main objective was to construct an Experimental setup with which it can be experimented that increase in the surface area would have an effect on the rate of heat transfer. Heat transfer through a circular pipe is investigated while using different geometries inside the pipe. The material of the pipe used is Stainless Steel and the working fluid is water. 4 pipes in a series were used in the experiment and each of them was 4 feet (1.2192 meter) long and the inner and outer diameter of the pipes is 2 inch (0.0508 meter) and 2.016 inch (0.0512 meter) respectively. Three of the four pipes were insulated with glass wool. A spring and a wire mesh, both of stainless steel material, was inserted in two insulated pipes. Then hot water with different temperature was flown through the pipes. The experiments were conducted for five different flow rates of the water and for each flow rate; five inlet temperatures of the water were taken. The temperature readings were taken at different locations of each pipe and calculated the heat transfer. The results are compared with other published papers. The results show that the internal surface area, fluid flow rate and even the initial fluid temperature play major role on heat transfer.