Authors:
Maryam Boumrah
1
;
Samir Garbaya
2
and
Amina Radgui
1
Affiliations:
1
Centre d’études doctorales Télécoms et Technologies de l’Information (CEDOC-2TI)”, INPT, Rabat, Morocco
;
2
Laboratoire END-ICAP, INSERM UMR1179, Arts et Metiers Institute of Technology, CNAM, LIFSE, HESAM University, F-75013 Paris, France
Keyword(s):
Fog Computing, Opentelemetry, Remote Patient Monitoring, Rehabilitation, Stroke, Latency, Scalability.
Abstract:
Remote rehabilitation of stroke patients reinforces in-person rehabilitation and enhances the regaining of neuromotor capabilities. However, monitoring stroke patients’ rehabilitation from different locations and on a large scale requires a low latency and scalable approach. A real-time visual analytics framework for monitoring in-home rehabilitation of stroke patients based on fog computing is proposed. The objective of this paper is to evaluate the performance of the proposed framework in terms of latency and scalability. OpenTelemetry was used for the evaluation of the proposed framework. OpenTelemetry was chosen over simulation tools for its real-time observability features providing accurate comprehension of the distributed system behaviors in real-world implementation. Five scenarios were setup by progressively escalating the volume of data flow and the number of packets. These scenarios enabled a thorough examination of the framework’s ability to handle higher workloads and sc
alability. The results of end-to-end latency of the proposed system were compared to the Cloud-only implementation. Compared to Cloud-only implementation, the findings of the evaluation showed that the latency of the proposed system was significantly low. Reflecting the scalability feature, the capacities of handling workload by the proposed system in terms of latency, throughput, processing, and resource utilization were stable across the first four configurations. However, the limitations noticed in the fifth configuration put in evidence the constraints of the experimental setup used in this research. Moreover, the scalability and efficiency of the system can be further enhanced in a distributed deployment in real-world conditions.
(More)