Sudarshan K. Srinivasan

William Won, Taekyung Heo, Saeed Rashidi et al.

As deep learning models and input data are scaling at an unprecedented rate, it is inevitable to move towards distributed training platforms to fit the model and increase training throughput. State-of-the-art approaches and techniques, such as wafer-scale nodes, multi-dimensional network topologies, disaggregated memory systems, and parallelization strategies, have been actively adopted by emerging distributed training systems. This results in a complex SW/HW co-design stack of distributed training, necessitating a modeling/simulation infrastructure for design-space exploration. In this paper, we extend the open-source ASTRA-sim infrastructure and endow it with the capabilities to model state-of-the-art and emerging distributed training models and platforms. More specifically, (i) we enable ASTRA-sim to support arbitrary model parallelization strategies via a graph-based training-loop implementation, (ii) we implement a parameterizable multi-dimensional heterogeneous topology generation infrastructure with analytical performance estimates enabling simulating target systems at scale, and (iii) we enhance the memory system modeling to support accurate modeling of in-network collective communication and disaggregated memory systems. With such capabilities, we run comprehensive case studies targeting emerging distributed models and platforms. This infrastructure lets system designers swiftly traverse the complex co-design stack and give meaningful insights when designing and deploying distributed training platforms at scale.

Zeyad A. Al-Odat, Sudarshan K. Srinivasan, Eman M. Al-Qtiemat et al.

This paper introduces a reliable health care system for diabetic patients based on the Internet of Things technology. A diabetic health care system with a hardware implementation is presented. The proposed work employs Alaris 8100 infusion pump, Keil LPC-1768 board, and IoT-cloud to monitor the diabetic patients. The security of diabetic data over the cloud and the communication channel between health care system components are considered as part of the main contributions of this work. Moreover, an easy way to control and monitor the diabetic insulin pump is implemented. The \mbox{patient\textquotesingle s} records are stored in the cloud using the Keil board that is connected to the infusion pump. The reliability of the proposed scheme is accomplished by testing the system for five performance characteristics (availability, confidentiality, integrity, authentication, and authorization). The Kiel board is embedded with Ethernet port and Cortex-M3 micro-controller that controls the insulin infusion pump. The secure hash algorithm and secure socket shell are employed to achieve the reliability components of the proposed scheme. The results show that the proposed design is reliable, secure and authentic according to different test experiments and a case study of the Markov model. Moreover, a 99.3\% availability probability has been achieved after analyzing the case study.

Zeyad A. Al-Odat, Sudarshan K. Srinivasan, Eman Al-qtiemat et al.

This paper introduces an Internet of Things (IoT)-based data acquisition and monitoring scheme for insulin pumps. The proposed work employs embedded system hardware (Keil LPC1768-board) for data acquisition and monitoring. The hardware is used as an abstract layer between the insulin pump and the cloud. Diabetes data are secured before they are sent to the cloud for storage. Each patient's record is digitally signed using a secure hash algorithm mechanism. The proposed work will protect the patient's records from being breached from unauthorized entities, and authenticates them from improper modifications. The design is tested and verified using $μ$Vision studio, the Keil board mentioned above, and an ALARIS 8100 infusion pump. Moreover, a test case for a real cloud example is presented with the help of the Center of Computationally Assisted System and Technology. This center provided the infrastructure service to test our work.