Communication-Aware Dynamic Edge Computing (CONNECT)
Lead Research Organisation:
Imperial College London
Department Name: Electrical and Electronic Engineering
Abstract
Internet of things (IoT) is slowly permeating every aspect of our lives; however, we are far from having a truly intelligent IoT. Smart sensors generate massive amounts of data continuously; for instance, an autonomous vehicle is expected to generate about one gigabyte of data per second, but more often than not data is not systematically processed, stored, or analyzed for better inference. Many specialized machine learning (ML) algorithms have been developed to learn from sensor measurements, but these assume a centralized setting, where data is available at a central processor with powerful computation capabilities. This centralized approach assumes that the massive amount of sensor data is transmitted to a cloud center, which may not be feasible due to limitations of the devices and channels, not meet the stringent delay constraints of most applications, e.g., controlling an autonomous vehicle, or the privacy requirements of users. In the CONNECT project, our goal is to develop real edge intelligence by enabling edge nodes to make local decisions rapidly and reliably in a collaborative manner. This will be achieved by developing novel caching, distributed computing and networking methodologies to enable federated/ distributed learning taking into account the network dynamics and physical channel variations. The developed joint computing, caching and communication framework will then be applied to a hierarchical heterogeneous architecture for vehicular ad-hoc networks (VANETs). This will not only enable efficient and reliable learning across mobile nodes, but also improve the security and privacy of autonomous cars by limiting decision making to local neighborhood. Integration of caching, computing and networking will be demonstrated both through large-scale simulations, and on a small-scale implementation platform, consisting of two cars and a roadside unit at Koc University. This project is expected to enable many data intensive edge applications, from multimedia content streaming to participatory data collection in mobile networks, including autonomous cars, drones, mobile robots and mobile cellular users.
Planned Impact
N/I
People |
ORCID iD |
| Deniz Gunduz (Principal Investigator) |
Publications
Amiri M
(2022)
Convergence of Federated Learning Over a Noisy Downlink
in IEEE Transactions on Wireless Communications
Amiri M
(2021)
Blind Federated Edge Learning
in IEEE Transactions on Wireless Communications
Amiri M
(2021)
Convergence of Update Aware Device Scheduling for Federated Learning at the Wireless Edge
in IEEE Transactions on Wireless Communications
Buyukates B
(2021)
Gradient Coding with Dynamic Clustering for Straggler Mitigation
Buyukates B
(2023)
Gradient Coding With Dynamic Clustering for Straggler-Tolerant Distributed Learning
in IEEE Transactions on Communications
Ceran E
(2021)
A Reinforcement Learning Approach to Age of Information in Multi-User Networks With HARQ
in IEEE Journal on Selected Areas in Communications
Chen M
(2021)
Distributed Learning in Wireless Networks: Recent Progress and Future Challenges
in IEEE Journal on Selected Areas in Communications
Gunduz D
(2023)
Beyond Transmitting Bits: Context, Semantics, and Task-Oriented Communications
in IEEE Journal on Selected Areas in Communications
Hasircioglu B
(2021)
Speeding Up Private Distributed Matrix Multiplication via Bivariate Polynomial Codes
| Description | This is a multi-national project in collaboration with partners from University of Oulu (Finland) and Koc University (Turkey). So far, the achievements of the Imperial team can be summarised as follows: We have worked towards improving the performance and efficiency of distributed learning performance over noisy and bandlimited wireless links. We have also addressed the problem of age-of-information, when status updates are conveyed over error-prone wireless channels. This is then adopted for the delivery of model updates in distributed learning. We have also developed reliable and secure distributed computation algorithms that will lay the foundations of secure distributed machine learning applications. We have also addressed the critical privacy problem in federated learning. Finally, we have worked towards implementing federated learning for mmWave beam selection in vehicular networks. In particular, we have developed an efficient federated learning solution for LIDAR-aided beamforming that can reduce the communication load of training significantly. We have also made advances for efficient content storage and delivery over unreliable wireless channels, which can be particularly attractive for vehicular networks. |
| Exploitation Route | This is a collaborative project, and our partner Koc University has an autonomous driving testbed consisting of several electric cars. One of the main objectives of the project is to implement some of the algorithms developed by Imperial on this testbed. Once their efficacy is proven, we are planning to contact autonomous driving companies to explore the adoption of these technologies for commercial use. There has been a bit limited progress in this implementation aspect of the project as the members of the Imperial and Oulu teams could not visit Koc University due to Covid, which prevented the collaborative work on the implementation of the federated learning algorithms developed by the Imperial and Oulu teams on the vehicular communication network testbed at Koc University. |
| Sectors | Digital/Communication/Information Technologies (including Software) Transport |
| URL | https://connect.ku.edu.tr |
| Description | CONNECT CHIST-ERA joint project |
| Organisation | Koc University |
| Country | Turkey |
| Sector | Academic/University |
| PI Contribution | CONNECT is a collaborative project funded under the CHIST-ERA program of European Union. In this collaboration, the particular role of Koc University is to implement the distributed machine learning algorithms developed by Imperial in their autonomous vehicle testbed. |
| Collaborator Contribution | Koc University provides access to their autonomous vehicle testbed as well as to data collected from the testbed. University of Oulu will collaborate with Imperial in the development of efficient edge learning solutions. |
| Impact | This is an interdisciplinary collaboration as Imperial focuses more on the machine learning and communication theoretic aspects while Koc University will focus on their implementation on vehicular networks and Oulu will focus on wireless communications and networking aspects. There is no outcome to report on yet (the project is in its first year), but joint research activities are ongoing. |
| Start Year | 2020 |
| Description | CONNECT CHIST-ERA joint project |
| Organisation | University of Oulu |
| Country | Finland |
| Sector | Academic/University |
| PI Contribution | CONNECT is a collaborative project funded under the CHIST-ERA program of European Union. In this collaboration, the particular role of Koc University is to implement the distributed machine learning algorithms developed by Imperial in their autonomous vehicle testbed. |
| Collaborator Contribution | Koc University provides access to their autonomous vehicle testbed as well as to data collected from the testbed. University of Oulu will collaborate with Imperial in the development of efficient edge learning solutions. |
| Impact | This is an interdisciplinary collaboration as Imperial focuses more on the machine learning and communication theoretic aspects while Koc University will focus on their implementation on vehicular networks and Oulu will focus on wireless communications and networking aspects. There is no outcome to report on yet (the project is in its first year), but joint research activities are ongoing. |
| Start Year | 2020 |