Project PICCOLO

Abstract

Piccolo will offer foundational technology breakthroughs for the long-term evolution of mobile networks, including 5G and beyond, enabling those networks and their operators as well as new market entrants to support newly emerging applications and services – and thus also foster broad innovation. As key technology, Piccolo will introduce an unprecedented blending of networking, computing and data storage capabilities in a uniform in-network computing architecture. This platform will allow flexible, easy, and impromptu instantiation of applications across network nodes, close to the user, by combining lambda functions using lightweight and secure virtualization techniques. The instant and ephemeral nature of the Piccolo approach will nicely complement (and be able to interact with) the presently standardised mobile edge computing and its service creation model.

Piccolo will drive its development from two demanding use cases related to privacy-preserving data processing: 1) in the automotive domain for blending in-car and edge computing for data processing and control loops; and 2) in the area of Internet of Things for IoT vision processing and associated AI.

Piccolo will develop new solutions for in-network computing that remove known and emerging deficiencies of edge/fog computing. Our vision is that every node includes processing, storage and networking capabilities in an integral architecture. The technical focus is twofold: (1) Compute platforms for network and 3rd party functions addressing fast, lightweight, secure virtualisation and data plane isolation ; (2) Distributed systems to ensure the network is scalable, resilient and secure, and supporting the joint optimisation of compute, network, and storage resources.

Our objective is to create an Internet principle of “transparent in-network computing” that complements today’s “transparent networking” and so enables a fresh dimension for permissionless innovation and growth.

Piccolo will deliver new technical concepts, architecture and mechanisms, underpinned by an understanding of the use cases for in-network computing. Those will be realized in (open) software systems, extensively evaluated in several use cases, and disseminated via software, (pre-)standardization efforts, and, ultimately, enhanced products of the industry partners.