Network for Distributed Intelligence: a Survey and Future Perspectives

Abstract

To keep pace with the explosive growth of Artificial Intelligence (AI) and Machine Learning (ML)-dominated applications, distributed intelligence solutions are gaining momentum, which exploit cloud facilities, edge nodes and end-devices to increase the overall computational power, meet application requirements, and optimize performance. Despite the benefits in terms of data privacy and efficient usage of resources, distributing intelligence throughout the cloud-to-things continuum raises unprecedented challenges to the network design. Distributed AI/ML components need high-bandwidth, low-latency connectivity to execute learning and inference tasks, while ensuring high-accuracy and energy-efficiency. This paper aims to explore the new challenging distributed intelligence scenario by extensively and critically scanning the main research achievements in the literature. In addition, starting from them, the main building blocks of a network ecosystem that can enable distributed intelligence are identified and the authors' views are dissected to provide guidelines for the design of a ''future network for distributed Intelligence''.

Distributed Machine Learning and Native AI Enablers for End-to-End Resources Management in 6G

“…Here, our aim is to concentrate on a subset of methodologies that are more pertinent to the 6G vision, in alignment with the challenges discussed in the previous section. Specifically, using as a starting point existing surveys, we examine the targeted subset of methodologies and assess their suitability for 6G in terms of the following benchmarking criteria: (a) Whether they are scalable to larger topologies [1] (scalability); (b) Whether they can become a part of a larger end-to-end framework for RM [4,14,16] (composability/modularity); (c) Whether their accumulated knowledge can be extracted and reused [11] (transferability).…”

“…However, neither the interactions between the operating infrastructure and a knowledge reuse mechanism nor other AI enablers are discussed. Works [14,15] provide reviews of end-to-end distributed intelligence from the "network-for-AI" viewpoint, targeting next-generation networks. This aspect of distributed intelligence focuses on how the network technologies can support AI applications running on the network, as opposed to the "AI-for-network" viewpoint, adopted here, that focuses on how AI can be leveraged for managing the network itself.…”

Survey of Novel Architectures for Energy Efficient High-Performance Mobile Computing Platforms

“…In this way, the computational performance and bandwidth required for each of the edge computing servers is greatly reduced, decreasing the overall cost and complexity of the system. This technique has gained traction in recent years, with much research being conducted to find the most efficient way to distribute a workload across an edge computing environment and different network typologies to yield the fastest and most reliable results [8][9][10]. While edge computing does offer several advantages over the more traditional, centralized data processing, it still does not fully address the concerns outlined in the previous paragraph.…”