hands holding iPhone

What We Do

Edge computing is driven by the enormous growth in data collected by billions of internet-of-things and mobile devices. Instead of sending data to the cloud for processing and storage, edge computing uses a distributed model where the processing is done much closer to where the data is created. This offers the potential for faster, near real-time response, lower cost, improved security, and greater power efficiency. However, such a computing architecture relies critically on reliable networks and data-driven AI techniques to operate at scale.

Athena’s edge computing development is driven by three insights:

  1. Recent advances in AI provide powerful tools to develop new data-driven approaches for networked computing system design, operation, and applications at scale.
  2. New AI techniques are required to handle the extreme heterogeneity and volume of client data generated by diverse and mobile edge client hardware, i.e. internet-of-things and wireless devices.
  3. Mobile networks can be leveraged by machine learning modalities involving collaboration amongst clients (federated learning), edge datacenters, and the Cloud (distributed learning) to offer new services and enhanced trustworthiness.

Our AI research for edge computing is organized under four inter-related thrusts:

Thrust I: Advancing AI for mobile networks develops AI techniques that fulfill the needs of the next-generation mobile network to guarantee functionality, efficiency, and trustworthiness. This thrust also exploit new AI computing frameworks and hierarchy which can adapt to the heterogeneous networks.

Thrust II: AI-powered computer systems at the edge designs next-generation edge data-centers with high efficiency, availability, and security for cloud-based mobile networks. This thrust also investigates systems support for AI at the edge, and develops AI-power control planes for mobile network infrastructures.

Thrust III: AI-powered networking at the edge creates new adaptable, scalable, and performance- aware mobile network infrastructure to provide flexible services for heterogeneous use cases. This thrust also explore s data-driven approaches to complement and augment existing algorithmic alternatives in network design.

Thrust IV: Services and applications develops innovative services and applications for the next- generation mobile networks enabled and inspired by AI technology, including assured, robust, and resilient services that facilitate design of autonomous systems at the edge.

Participating Universities

Duke University logo
MIT logo
NC A&T logo
Princeton University logo
University of Michigan logo
University of Wisconsin logo
Yale logo