Sustainable Computing and Communication at the Edge (SONATA)

Lead Research Organisation: Imperial College London
Department Name: Electrical and Electronic Engineering

Abstract

Modern communication networks are rapidly evolving into sophisticated systems combining communication and computing capabilities. Computation at the network edge is the key to supporting many emerging applications, from extended reality to smart health, smart cities, smart factories and autonomous driving. SONATA is motivated by the fact that the large scale adoption of edge intelligence technology, while benefiting human productivity and efficiency, will result in a surge of data and computation in mobile networks, which, in turn, will exacerbate their already significant energy consumption. SONATA is an interdisciplinary effort to tame this growing energy demand by combining memristive hardware and energy harvesting technologies with novel machine learning algorithms and physical layer communication techniques. In particular, we want to combine the energy efficient in-memory computing and learning potential of memristive devices with an "over-the-air computation (OAC)" approach to edge learning, which turns the air from a purely communication medium to a computation unit. Our project not only aims at reducing the energy requirements of edge learning systems drastically, but also focuses on making them robust against stochastic failures, due to unreliable hardware or energy sources. We will exploit tools from circuit design, coding theory, wireless communications, machine learning and network science to achieve these goals. Results from SONATA will open up new directions for research and development of technologies that will allow mobile systems to offer the much anticipated communication and computing services in a sustainable manner.

Publications

10 25 50

publication icon
Gunduz D (2023) Beyond Transmitting Bits: Context, Semantics, and Task-Oriented Communications in IEEE Journal on Selected Areas in Communications

publication icon
Lo W (2023) Collaborative Semantic Communication for Edge Inference in IEEE Wireless Communications Letters

publication icon
Ozfatura E (2022) All You Need Is Feedback: Communication With Block Attention Feedback Codes in IEEE Journal on Selected Areas in Information Theory

publication icon
Shao Y (2023) Semantic Communications With Discrete-Time Analog Transmission: A PAPR Perspective in IEEE Wireless Communications Letters

publication icon
Shao Y (2023) Bayesian Over-the-Air Computation in IEEE Journal on Selected Areas in Communications

publication icon
Wu H (2022) Channel-Adaptive Wireless Image Transmission With OFDM in IEEE Wireless Communications Letters

 
Description This is an ongoing research project. In this period of the project, we have analysed a real drift dataset for titanium dioxide memristive devices, consisting of 2751 cumulative device hours of readings taken from varying initial resistances for the devices. We proposed a simple model of the formation of a filament in the titanium dioxide memristor to explain the switching mechanism and linear dependence of the conductance on the proposed state variable, introduced in the event-based modelling framework previously developed. We fit the event-based model to the dataset as a demonstration of how the event-based modelling approach could be aligned to dataset statistics. We improved modelling techniques previously developed for delay-aware storage using autoencoders, to allow for improved performance across a wide range of delays.

In joint work with Pázmány University, Budpest, we explored the state characterisation of commercially available self directed channel memristive devices. We based the state characterisation on exponential modelling approaches, demonstrating a good fit to the data collected. Ongoing and future work aims to characterise the energy-information trade-off of such devices through resistive drift measurements and its potential implications for the storage of different kinds of information on memristive devices, for both traditional storage and neuromorphic computing.
Exploitation Route At this stage, the research is on very low TRL level; therefore, we mainly expect academic impact; however, there is significant potential of memristive devices to be used in energy-efficient implementation of neural networks. The ideas developed in this project can be taken forward by the industry to develop practical memristive computing systems.
Sectors Digital/Communication/Information Technologies (including Software)

Electronics

URL https://www.chistera.eu/projects/sonata