Amarjit Singh Dhillon who is working as a research assistant at Carleton University, Canada, greatly contributed in a research project focused on devising new techniques for edge computing-based Complex Event Processing (CEP) for sensor-based systems. His research led to the development of a Mobile Complex Event Processing (MCEP) system that uses an edge-based computing technique which can be used in various use cases such as Remote Patient Monitoring (RPM), smart homes, and smart cities. The proof-of-concept prototype for the proposed system is designed to detect complex events on the edge device which can be a mobile device (e.g. a smart phone) or a Raspberry-Pi board for example. Amarjit has validated the system prototype by considering an RPM-based use case. The RPM technique is widely used in remote healthcare where the patient is remotely monitored for various issues such as Arrhythmia, and Congestive Heart Failure (CHF) without being admitted to a hospital. In case of remote patient monitoring, a complex event represents a health-threatening event for a patient, which results from the combination of multiple raw events each of which may correspond for example to the output of a sensor detecting an excessive increase in heart rate and an abnormal increase in respiration rate within some time interval.
In the current state-of-the-art centralized server-based techniques used for RPM, the sensor streams (representing the raw events) are collected on the patient’s mobile edge device through wireless/wired non-invasive sensors which are deployed on the patient’s body. These sensor data streams are then forwarded to the back-end hospital server, where processing is done using service, such as the CEP-as-a-Service (CEPaaS), to detect the complex events. According to Mr. Dhillon, the biggest disadvantage of this centralized system is that the patient’s mobile phone has to be remain connected to the hospital server at all times in order for the sensor data streams to reach the remote centralized server. For example, in the case of remote areas where internet connectivity is not available, the patient’s mobile device will have no network connection. As a repercussion of this, if a health-threatening event takes place, neither the patient nor the hospital will be able to detect that event because there is no network, meaning that the sensor data streams are not forwarded to the Hospital server. In such a scenario, the patient will not receive medical care in a timely fashion which may further worsen the patient’s condition. Hence, the server-based CEP system is not efficient in rural/ remote areas where the network is not accessible at all times. Such a centralized system will also lead to increased user cost, grater queuing delays on a hospital server which serves multiple patients. Also, the mobile device power consumption will be increased due to large volumes of data transfer between the mobile device and the hospital server.
Amarjit Dhillon worked under the supervision of Professor Shikharesh Majumdar and Professor Marc St Hilaire in order to resolve the aforementioned issues, and Ali El-Haraki was the collaborator from TELUS. This research received funding from a TELUS Research Grant (Principal Investigator: Dr. Majumdar) for the “Stream Processing Service“ project. Mr. Dhillon proposed a new mobile device-based CEP system which uses an embedded CEP engine on an Android-based mobile device that detects complex events and sends them to an Apache-licensed WSO2 Internet of Things (IoT) server, which is used as the back-end hospital server for further processing. Since the application on the mobile device detects the complex events, in a scenario with network unavailability, this RPM system will alarm the patient if there is something wrong with him/her. The proof-of-concept prototype for the novel MCEP system that Mr. Dhillon proposed has four components as explained next. (1) A mobile device that collects the patient’s data and has the application to detect the complex events, (2) a workstation which contains sensor simulators, (3) the IoT Hospital Server (IHS) and, (4) a wireless router that helps connect various components such as the mobile device, IHS and the sensor simulator. The proposed system also consumes less battery power in comparison to the centralized server-based system. For the experiments conducted on both the proposed and the centralized system, the proposed system resulted in the savings of $13/hour as per the current network usage rate charged by BELL, which is one of the largest telecommunication providers in Canada.
A recent conference paper entitled, “MCEP: A Mobile Device Based Complex Event Processing System for Remote Healthcare”, which presented the proof-of-the-concept prototype system has won the IEEE Best Paper Award in the 11th IEEE International Conference on Internet of Things held at Halifax, Canada in July 2018. Another conference paper “A Mobile Complex Event Processing System for Remote Patient Monitoring” has been published in IEEE International Conference on the Internet of Things (ICIOT) held in San Francisco, in July 2018. The proposed system has also received a 3rd place at Data Day 5.0 student poster competition event conducted at Carleton University.
This novel edge-based MCEP engine was realized as a joint effort of Mr. Amarjit Singh Dhillon, his supervisors (Dr. Shikharesh Majumdar and Dr. Marc St Hilaire), and Mr. Ali El- Haraki (a TELUS representative).