Fog networking is an end-to-end horizontal architecture that distributes computing, storage, control, and network functions close to users along the cloud-to-thing continuum. This fog architecture includes the cloud, core, metro, edge, clients, and things. The fog architecture will further enable pooling, orchestrating, managing, and securing the resources and functions that are distributed in the cloud, anywhere along the cloud-to-thing continuum, and on the things or edge devices to support end-to-end services and applications. A fog architecture can bring significant benefits compared to a pure client- or server-based approach: for instance, putting some control at the client makes it easier to customize the application’s configuration according to individual clients’ preferences. Client-based control can also allow these configurations to more rapidly adapt to changes in client preferences or background information. A central server component, on the other hand, can improve coordination between different clients who might compete for resources, e.g., a cellular network setting in which multiple applications share network bandwidth. Thus, fog networking allows applications to optimize over both client and server information, leading to an intelligent division of labor between the client and the server.

This project has the following research thrusts and outcomes:

(1) Computation/processing decomposition framework. The existing distributed computing/processing frameworks in the cloud, however, are not directly applicable to fog; the cloud has mostly homogeneous computing nodes, a well-structured network topology, and reliable network connectivity, while the fog has to deal with other extremes, such as heterogeneous computing nodes with high churn rates and unstructured networks with frequent disruptions. To address this challenge, we developed Crystal, a simple, loosely coupled, distributed computing framework in Golang, which provides an easy abstraction for fog application development while supporting location-transparency, self-healing, auto-scaling, and mobility support. We open-sourced Crystal and used it for CS Master Students’ Independent Studies.

(2) Fog-Cloud Interface: While fog envisions a larger role for semi-independent client/edge devices, data centers, and the cellular core are sometimes still better for global coordination. We approach this research thrust in the context of mobile video streaming, and propose FLARE, a coordinated HTTP rate adaptation approach that incorporates both client- and network-side information and guarantees coordination between network- and client-side bitrate selection.

(3) From Local to Global: Can local visibility, decisions, and interactions lead to a desirable global state across the network? We developed CASTLE, a client-side network inference system that estimates the load of a given cellular network. CASTLE can infer the network load locally with minimal help from the core network and leverage this information to reduce the energy consumption of mobile devices, or selectively preload mobile applications to improve network congestion in real-time. We developed the CASTLE SDK on Android and made it available to the public.

(4) Incentivizing client participation. While client functions are often more nimble and easier to evolve, individual clients may require additional incentives to participate in fog-based systems. If carefully designed, such incentives may even steer client actions towards ones that globally optimize the network. For this thrust, we developed T-Chain, a simple, distributed, but highly efficient fairness-enforcing incentive mechanism for cooperative computing. T-Chain enforces reciprocity between client contributions to prevent exploitation or free-riding by misbehaving or selfish clients. T-Chain is distributed and simple to implement, as no trusted third party is required to monitor or enforce the scheme, nor is there any reliance on reputation information or tokens.

(5) Using device redundancy to achieve resilience. The presence of many client/edge devices in fog computing may offer the “power/wisdom of the crowd” for enhanced availability. However, this requires sufficient independence of each device’s visibility and actions. We developed CYRUS (Client-defined privacY protected Reliable cloUd Service), a practical system that realizes this architecture. CYRUS ensures user privacy and reliability by scattering files into smaller pieces across multiple CSPs (Cloud Service Providers), so that no one CSP can read users’ data.

The outcomes listed above were published in major CS conferences and journals such as ACM MobiSys, ACM CoNEXT, EuroSys, IEEE ICDCS, IEEE INFOCOM, Elsevier Computer Networks, and IEEE Transactions on Networking. This project impacted the industry by founding a startup Myota Inc. Myota provides a data security management solution that stores critical data across multiple storage nodes, regardless of storage type, while dynamically transforming a user’s security infrastructure with artificial intelligence. This project also created an open-source 4G LTE evolved packet core (EPC) software, NextEPC, which controls and manages 4G LTE networks. NextEPC’s codebase consists of over 280,000 lines of code that implements the latest 3GPP LTE Release 14 and runs on major operating systems (e.g., Linux, FreeBSD, and MacOSX). One female Ph.D. student was supported under this project.

Last Modified: 12/02/2019
Modified by: Sangtae Ha