My research interests lie in the broader areas of systems and networking. My focused research topics include distributed computing applied to Cyber-physical systems (CPS), the Internet of Things (IoT), and big data and intelligent systems.
A few research efforts are listed below. For a full list, please see the publications page.
Edge AI
Edge AI refers to the deployment of artificial intelligence (AI) algorithms/models directly on devices at the edge of the network—such as smartphones, sensors, cameras, and other IoT devices—rather than solely relying on centralized cloud servers. This approach allows data to be processed locally, reducing latency, improving privacy, and enabling real-time decision-making. This paradigm is particularly useful in applications where speed and reliability are critical, such as autonomous vehicles, smart surveillance, industrial automation, and healthcare monitoring. By minimizing the need to send data to the cloud, it also helps conserve bandwidth and enhances data security. In addition to an AI model executing on end-user devices, Edge AI enables the models (or their parts) to be hosted on nearby servers closer to the devices (edge servers).
IoT (Internet of Things) and Smart Spaces
IoT-enabled spaces, also known as smart spaces, are the physical spaces that are instrumented with sensors and actuators. IoT technologies enable capturing data from the physical world (sensing), processing those data and generating actionable information to pass to the users and actuators to act. Our work on building IoT platforms addresses a diversifies set of problems, such as optimal resource allocation and utilization, efficient data collection, scheduling for data uploads, and energy-aware perpetual operations arising in the context of IoT deployments in real world communities.
Edge computing for data-driven systems
Data-driven systems tend to follow the so-called observe-analyze-act cycle where data generated by devices and sensors are sent to the cloud where the accumulated data gets analyzed and the results are delivered to the end-users to act. Edge computing aims at bringing computation closer to both data and users, as opposed to pushing all data to the cloud for all computation. Interesting problems are solved in two edge aspects: cloud-facing edge (edge between sensors and the cloud) and user-facing edge (edge between the cloud and users).
Mobile computing
Mobile phones possessed by average individuals enable ample opportunities of sensing through 3D accelerometers and inertia sensing, location (GPS), audio video sensing (microphones, cameras) yet having issues like limited processing capabilities and data bandwidth, metered/budgeted data plans, and energy constraints. Our work addresses some of these in various application domains.
DTN protocols and services for disaster response
DTN (Delay-tolerant Network) enables intermittent but functional networking when the regular communication is disrupted due to interruption to telco services and the Internet and wide-scale power outage (say, the aftermath of a Hurricane). DTNs utilize opportunistic device-to-device communication among the moving entities in the scene, such as the first responders, police cars, relief vehicles, Wifi routers at home/vehicles, and mobile phones carried by people. The communication enables data services such as coordinating relief and rescue operations, gathering first-hand situational data, and damage assessment.