Edge Cloud Systems

The context aware, on-demand and distributed nature of next generation applications designed for Mobile-Cloud and Internet of Things scenarios, puts high pressure on devices and infrastructure due to sheer amount of generated traffic while simultaneously requiring even lower response latencies for their interactive nature. The resource contraints in temrs of computation, storage, I/O extension, etc are inherent to the mobile and end-user devices and pose additional challenges to meet above demands. We are exploring the provisioning and use of emerging 'Edge Cloud' tier situated only 1 physical hop away from end user devices (e.g., traditional PCs, homeservers, Wi-Fi routers, Small Cells etc.), which can be provisioned seamlessly on-demand by the backend cloud or by end user devices in ad-hoc manner via device discovery.

The goals of Edge Cloud project are to explore different system level approaches to (i) address challenges in realizing the 'edge cloud' tier, (ii) prototype novel edge cloud applications to showcase new functionality that can be enabled by it and (iii) explore ways to leverage it to directly improve existing applications on end user devices and indirectly benefit large scale applications running in the cloud data center by providing faster and/or aggregated ingress to user generated traffic.

Current Members

Associated Faculty



Recent Publications

  • Bhardwaj K., Shih M., Agarwal P., Gavrilovska A., Kim T., Schwan K., Fast, scalable and secure onloading of edge functions using AirBox, The First IEEE/ACM Symposium on Edge Computing 2016. (To appear)
  • Jang M., Lee H., Schwan, K., Bhardwaj, K., SOUL: An Edge cloud system for mobile applications in a sensor-rich world, The First IEEE/ACM Symposium on Edge Computing 2016. pdf
  • Bhardwaj K., Gavrilovska A., Schwan K., Ephemeral Apps, The Seventeenth Workshop on Mobile Computing Systems and Applications (ACM HotMobile 2016)
  • Bhardwaj K., Agarwal P.,Gavrilovska A., Schwan K., AppSachet: Distributed app delivery from the edge cloud, The 7th EAI International Conference on Mobile Computing, Applications and Services, 2015 (MobiCASE'15) 
  • Bhardwaj K., Agarwal P., Gavrilovska A., Schwan K., Allred, A., AppFlux: Taming App Delivery via Streaming, Conference on Timely Results in Operating Systems 2015 (TRIOS'15). Held in conjunction with SOSP'15. 
  • Jang M.; Lee H.; Bhardwaj, K.;Schwan, K., vSensor: Toward Sensor-rich Mobile Applications, Sensors to Cloud Architectures Workshop (SCAW-2015) Held in conjunction with HPCA 15.pdf
  • Minsung Jang; Schwan, K.; Bhardwaj, K.; Gavrilovska, A.; Avasthi, A., Personal clouds: Sharing and integrating networked resources to enhance end user experiences, INFOCOM, 2014 Proceedings IEEE , vol., no., pp.2220,2228, April 27 2014-May 2 2014 pdf
  • Bhardwaj, K.; Sreepathy, S.; Gavrilovska, A.; Schwan, K., ECC: Edge Cloud Composites, Mobile Cloud Computing, Services, and Engineering, 2014 2nd IEEE International Conference on Mobile Cloud, vol., no., pp.38,47, 8-11 April 2014 pdf

Current Projects:

  • AirBox is a flexible software platform for mobile edge cloud computing. It enables fast, secure provisioning and execution of services running on edge cloud platforms.
  • App*: AppFlux, AppSachets and AppCraft are novel edge services leveraging edge cloud infrastructure to benefit Android app ecosystem in terms of faster app delivery, lower congestion due to app traffic in the last mile and aiding in barriers to app discovery.
  • Virtual Sensor abstractions provide such applications with ease of use for today’s multitude of sensors. vSensors not only assist with sensor use, but also interact with nearby and cloud resources to make available to applications the resources required for processing sensor-acquired data and controlling actuators, thus permitting them to scale in sensor usage. 

Past Projects:  

  • Personal Cloud introduces a new runtime layer working on top of this improved Stratus. Unlike Stratus, Pcloud uses the service virtualization concept, so an application is now a set of services connected via API. Moreover, the runtime is responsible to allocate resources  for virtual platforms to run each service of an application. Since Pcloud deals with cloud resources, it monitors trade-off between local and cloud in terms of latency.
  • ECC dynamic virtual computational platforms.
  • Stratus by creating a virtual platform consisting of uniformly accessible local and remote resources granted by a runtime access control, applications can obtain higher levels of performance, improved ease-of-use, and fidelity going beyond the limitations of individual devices.

Thanks to our sponsors: NSF, VMWare, Intel ISTC-Embedded Computing.