[{"_id":"project:5996","_type":"project","abstract":{"en":"Networked sensing systems, such as¬†the Internet-of-Things (IoT), are revolutionizing many fields in industry, transportation, and healthcare for sensing and actuation. These systems connect numerous sensing devices, which have limited computation power, batteries, and user interfaces. Configuration and maintenance are difficult, and cost a lot of time and money. Automated and scalable system management is desirable for cost-effectiveness and safety.The project aims to provide a generic, scalable, and new¬†methodology to coordinate heterogeneous devices and reconfigure the system automatically in changing environments. (i) It develops novel and lightweight data analytics and machine learning algorithms for resource-limited sensing devices to enable intelligent functionality. (ii) It derives¬†an adaptive framework for coordinating heterogeneous devices in a distributed network to improve system performance. (iii) It provides system resilience by detecting abnormal behaviors and enabling auto-configuration to recover from failures and malicious attacks.The project will¬†be¬†conducted by Dr. Edith Ngai and a¬†PhD student. It will benefit from our research environment and the GreenIoT testbed that we developed in the city center of Uppsala. Our solution is expected to significantly reduce the management cost and improve resilience of networked sensing systems against failures and malicious attacks."},"project_id":"2017-04543_VR","identifier_short":"2017-04543","dates":{"start_date":"2018-01-01","end_date":"2021-12-31"},"organizations":[{"funding":[{"_id":2,"id":"202100-5208","sv":"Vetenskapsrådet","en":"Swedish Research Council"}]},{"coordinating":[{"_id":978,"id":"202100-2932","sv":"Uppsala universitet","en":"Uppsala University"}]}],"people":[{"project_leaders":[{"name":"Ngai, Edith","role":"principal_investigator","affiliation":[{"_id":978,"id":"202100-2932","sv":"Uppsala universitet","en":"Uppsala University"}]}]},{"other_personnel":[]}],"tags":[{"_id":11583,"id":"20206","sv":"Datorsystem","en":"Computer Systems"}],"titles":{"sv":"Adaptiva och robusta underhåll för nätverkssensorsystem","en":"Towards self-adaptive and resilient networked sensing systems"},"total_funding":"3520000","type_of_awards":{"sv":"Projektbidrag","en":"Project grant"},"publications":[{"id":"diva2:1727796","type":"article-journal","status":"Published","issued":{"date-parts":[[2024]]},"title":"An Experimental Study of Byzantine-Robust Aggregation Schemes in Federated Learning","language":"eng","author":[{"family":"Li","given":"Shenghui","ORCID":"0000-0003-0145-3127","localId":"sheli564","affiliation":[{"id":"872701","name":"Uppsala universitet, Datorteknik"},{"id":"1107","name":"Uppsala universitet, Avdelningen för datorteknik"}]},{"family":"Ngai","given":"Edith C.-H.","ORCID":"0000-0002-3454-8731","localId":"eding810","affiliation":[{"name":"Department of Electrical and Electronic Engineering, The University of Hong Kong, Hong Kong, China"}]},{"family":"Voigt","given":"Thiemo","ORCID":"0000-0002-2586-8573","localId":"thvoi765","affiliation":[{"id":"888150","name":"Uppsala universitet, Nätverksbaserade inbyggda system"},{"name":"RISE, the Research Institutes of Sweden, Stockholm, Sweden"}]}],"abstract":"Byzantine-robust federated learning aims at mitigating Byzantine failures during the federated training process, where malicious participants (known as Byzantine clients) may upload arbitrary local updates to the central server in order to degrade the performance of the global model. In recent years, several robust aggregation schemes have been proposed to defend against malicious updates from Byzantine clients and improve the robustness of federated learning. These solutions were claimed to be Byzantine-robust, under certain assumptions. Other than that, new attack strategies are emerging, striving to circumvent the defense schemes. However, there is a lack of systematical comparison and empirical study thereof. In this paper, we conduct an experimental study of Byzantine-robust aggregation schemes under different attacks using two popular algorithms in federated learning, FedSGD and FedAvg . We first survey existing Byzantine attack strategies, as well as Byzantine-robust aggregation schemes that aim to defend against Byzantine attacks. We also propose a new scheme, ClippedClustering, to enhance the robustness of a clustering-based scheme by automatically clipping the updates. Then we provide an experimental evaluation of eight aggregation schemes in the scenario of five different Byzantine attacks. Our experimental results show that these aggregation schemes sustain relatively high accuracy in some cases, but they are not effective in all cases. In particular, our proposed ClippedClustering successfully defends against most attacks under independent and identically distributed (IID) local datasets. However, when the local datasets are Non-IID, the performance of all the aggregation schemes significantly decreases. With Non-IID data, some of these aggregation schemes fail even in the complete absence of Byzantine clients. Based on our experimental study, we conclude that the robustness of all the aggregation schemes is limited, highlighting the need for new defense strategies, in particular for Non-IID datasets.","DOI":"10.1109/tbdata.2023.3237397","ScopusId":"2-s2.0-85147301735","NBN":"urn:nbn:se:uu:diva-494317","issue":"6","volume":"10","page":"975-988","container-title":"IEEE Transactions on Big Data","ISSN":"2332-7790","keyword":"Byzantine attacks; distributed learning; federated learning; neural networks; robustness","publisher":"Institute of Electrical and Electronics Engineers (IEEE)","published":[{"raw":"2023-01-17T11:38:00.000+01:00"}],"created":[{"raw":"2023-01-17T11:38:21.762+01:00"}],"updated":[{"raw":"2025-02-19T16:20:50.247+01:00"}],"URL":"https://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-494317"},{"id":"diva2:1895304","type":"paper-conference","issued":{"date-parts":[[2024]]},"title":"Blades : A Unified Benchmark Suite for Byzantine Attacks and Defenses in Federated Learning","language":"eng","author":[{"family":"Li","given":"Shenghui","ORCID":"0000-0003-0145-3127","localId":"sheli564","affiliation":[{"id":"1107","name":"Uppsala universitet, Avdelningen för datorteknik"},{"id":"872701","name":"Uppsala universitet, Datorteknik"}]},{"family":"Ngai","given":"Edith C. H.","affiliation":[{"name":"Univ Hong Kong, Hong Kong, Peoples R China"}]},{"family":"Ye","given":"Fanghua","affiliation":[{"name":"UCL, London, England"}]},{"family":"Ju","given":"Li","ORCID":"0000-0001-9500-1791","localId":"liaju272"},{"family":"Zhang","given":"Tianru","ORCID":"0000-0001-9983-3755","localId":"tiazh991","affiliation":[{"id":"9908","name":"Uppsala universitet, Avdelningen för beräkningsvetenskap"},{"id":"3400","name":"Uppsala universitet, Tillämpad beräkningsvetenskap"}]},{"family":"Voigt","given":"Thiemo","ORCID":"0000-0002-2586-8573","localId":"thvoi765","affiliation":[{"id":"888150","name":"Uppsala universitet, Nätverksbaserade inbyggda system"},{"name":"Res Inst Sweden, Stockholm, Sweden"}]}],"abstract":"Federated learning (FL) facilitates distributed training across different IoT and edge devices, safeguarding the privacy of their data. The inherent distributed structure of FL introduces vulnerabilities, especially from adversarial devices aiming to skew local updates to their advantage. Despite the plethora of research focusing on Byzantine-resilient FL, the academic community has yet to establish a comprehensive benchmark suite, pivotal for impartial assessment and comparison of different techniques. This paper presents Blades, a scalable, extensible, and easily configurable benchmark suite that supports researchers and developers in efficiently implementing and validating novel strategies against baseline algorithms in Byzantine-resilient FL. Blades contains built-in implementations of representative attack and defense strategies and offers a user-friendly interface that seamlessly integrates new ideas. Using Blades, we re-evaluate representative attacks and defenses on wide-ranging experimental configurations (approximately 1,500 trials in total). Through our extensive experiments, we gained new insights into FL robustness and highlighted previously overlooked limitations due to the absence of thorough evaluations and comparisons of baselines under various attack settings. We maintain the source code and documents at https://github.com/lishenghui/blades.","ISBN":"979-8-3503-7025-6","DOI":"10.1109/IoTDI61053.2024.00018","ScopusId":"2-s2.0-85196568437","NBN":"urn:nbn:se:uu:diva-537577","page":"158-169","container-title":"2024 IEEE/ACM Ninth International Conference on Internet-of-Things Design and Implementation (IoTDI)","event":"9th ACM/IEEE Conference on Internet of Things Design and Implementation (IoTDI), May 13-16, 2024, Hong Kong, Hong Kong","keyword":"Byzantine attacks; distributed learning; federated learning; IoT; neural networks; robustness","publisher":"Institute of Electrical and Electronics Engineers (IEEE)","published":[{"raw":"2024-09-05T11:41:00.000+02:00"}],"created":[{"raw":"2024-09-05T11:41:38.106+02:00"}],"updated":[{"raw":"2025-02-11T12:41:12.221+01:00"}],"URL":"https://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-537577"}],"links":[{"type":"pid","link":"https://uu.diva-portal.org/smash/api/project/swecris/project:5996"}]}]