Distributed storage and computing on the edge (of the network)
Today's mobile networks are populated by all sorts of devices which offer —in addition to a significant amount of sensing capabilities and myriad services to the user— a wealth of processing power, inexpensive storage and a wide range of computational and networking resources. Often times, the computation resources available on end-devices and other network elements remain highly underused. This inefficient use of resources is a consequence of the dominating cloud computing model in which the vast majority of online services and applications offload computationally expensive tasks to centralised and large scale cloud infrastructure services like Amazon EC2 even when they have computational resources available in their proximity.
As an alternative to the highly centralised cloud computing model, the “fog computing” paradigm has been recently proposed. This novel paradigm aims to leverage an ecosystem of computing resources distributed all over the communication devices at the edge of the network (e.g., base stations and CPEs), even considering end user devices (e.g.,smartphones) and resource-constrained devices (e.g., IoT sensors). As a result, fog computing aims to complement the existing cloud computing model by leveraging diverse and richer resources otherwise underused. Fog computing will enable unseen use cases, future applications and technological opportunities that the cloud computing model cannot provide.
This project will study opportunities, technologies, marketing strategies and policies to advance the fog computing paradigm. We will explore challenges so that end user devices, applications and access network devices can dynamically and securely share and access any computational resource (storage, networking, sensing and computing power) available in their vicinity (e.g., WiFi islands, home network deployments, trusted devices forming a personal or community cloud and even 5G-and-beyond mobile radio networks). However, in this project we will also explore economic and sociological challenges to guarantee user trust, fairness and security when accessing resources from third-party services. We will study and consider both scenarios in which all devices are trusted (e.g., family-based cloud) and others in which there may be potential adversaries, free-riders and malicious agents. Namely, we will analyse the feasibility and effectiveness of incentive-based models and reputation mechanisms to promote user participation and maximise fairness.
In this project we will analyse the need for a third-party authority and enabler, namely an Internet Service Provider (ISP). In fact, the ISP not only can offer a vast infrastructure located at the edge of the network (including CPEs and base stations), but it also has a position of power to guarantee fairness and provide trust. To realise this vision and in order to make it a win-win solution for all stakeholders, the system needs to meet a number of design goals: 1)it must support the configuration and setup of resource pools, 2) it must relay on light and distributed algorithms for the control of local connectivity (e.g., on D2D communication schemes and opportunistic, context-aware IoT protocols) and 3) it must run well-designed and secure mechanisms for accessing the resource pool. The whole system should be network-aware, taking into account the nature of the access technology and possible inefficiencies found in the last mile, and in the leveraged wireless technologies.
IMDEA Networks Institute