[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-120971-en":3,"doc-seo-120971-105":29,"detail-sidebar-cat-0-en-105":90},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":11,"language":21,"language_code":22,"site_id":23,"html_lang":22,"table_of_contents":24,"faqs":25,"seo_title":26,"seo_description":14,"update_tm":27,"read_time":28},120971,137441390410,"Hazel","https://ap-avatar.wpscdn.com/avatar/2000252f4ab5702993?_k=1776741390130283984",8,"Research & Report","IoTGAN - GAN Powered Camouflage Against Machine Learning Based IoT Device Identification","With the expansion of IoT deployments, machine-learning-driven device identification increasingly becomes a security prerequisite. Yet such methods depend heavily on training data and do not inherently resist adversarial manipulation. The proposed IoTGAN attack manipulates IoT traffic to bypass ML-based identification by addressing black-box substitute-model learning and traffic perturbation generation. Experimental results show IoTGAN can meet its evasion objectives while maintaining device functionality, and the work further presents efficient countermeasures to harden identification defenses.","IoTGAN: GAN Powered Camou􀀃age Against  \nMachine Learning Based IoT Device Identi􀀂cation  \nTao Hou†, Tao Wang‡, Zhuo Lu†, Yao Liu† and Yalin Sagduyu§†University of South Florida, Tampa, FL, USA, {taohou@, zhuolu@, [yliu@cse.](yliu@cse.}usf.edu)[}](yliu@cse.}usf.edu)[usf.edu](yliu@cse.}usf.edu)[ ](yliu@cse.}usf.edu)‡New Mexico State University, Las Cruces, NM, USA, [taow@nmsu.edu](taow@nmsu.edu)  \n§ Intelligent Automation Inc, Rockville, MD, USA, [ysagduyu@i-a-i.com](ysagduyu@i-a-i.com)  \narXiv :2201 .03281v2 [ cs .CR] 16 Dec 2023  \nAbstract—With the proliferation of IoT devices, researchers have developed a variety of IoT device identi􀀂cation methods with the assistance of machine learning. Nevertheless, the security of these identi􀀂cation methods mostly depends on collected training data. In this research, we propose a novel attack strategy named IoTGAN to manipulate an IoT device’s traf􀀂c such that it can evade machine learning based IoT device identi􀀂cation. In the development ofIoTGAN, we have two major technical challenges:(i) How to obtain the discriminative model in a black-box setting, and (ii) How to add perturbations to IoT traf􀀂c through the manipulative model, so as to evade the identi􀀂cation while not in􀀃uencing the functionality of IoT devices. To address these challenges, a neural network based substitute model is used to 􀀂t the target model in black-box settings, it works as a discriminative model in IoTGAN. A manipulative model is trained to add adversarial perturbations into the IoT device’s traf􀀂c to evade the substitute model. Experimental results show that IoTGAN can successfully achieve the attack goals. We also develop ef􀀂cient countermeasures to protect machine learning based IoT device identi􀀂cation from been undermined by IoTGAN.  \nIndex Terms—Internet of Things, Device Identi􀀂cation, IoT Security, Machine Learning, Generative Adversarial Network  \nI. INTRODUCTION  \nThe Internet of Things (IoT) refers to the network of physical devices that are embedded with sensors, chips, operating systems and other technologies, collecting and exchanging data over Internet [1] . The popularization of universal computer chips and the ubiquity of wireless networks enable the revolution to transform traditional devices into smart devices as a part of the IoT. These IoT devices can be extensively deployed for different purposes including consumer (e.g., smart home, health care), commerce (e.g., transportation, manufacturing, agriculture), and military (e.g., battle􀀂eld equipment, autonomous reconnaissance) . However, the heterogeneity of these devices also imposes security challenges to the management of IoT networks.  \nFor a network containing different kinds of IoT devices, it is vital to identify the type of each IoT device before applying 􀀂ne-graded security policies. Other than managing different kinds of IoT devices locally by the entity which the devices belong to, knowing the type of the device can enable a global management for security purpose from the level of the whole network, in order to permit or prohibit IoT device’s speci􀀂c behavior. For example, in a military base, the network should keep the geographical information con􀀂dential and forbid the surveillance camera transferring video data to the outside.  \nAnother example is that an organization may have different permissions for its personnels to access different smart devices (e.g., the maintenance staff can adjust the air conditioner unit; the security guard can view the monitor video; and any person should be able to control the smart bubble). More importantly, IoT device identi􀀂cation can facilitate detecting vulnerable IoT devices and preventing malicious rogue IoT devices.  \nResearchers have proposed various methods for IoT device identi􀀂cation. A simple way is using identi􀀂ers (e.g., MAC addresses, IP addresses, Bluetooth ID, Zigbee ID) to identify IoT devices. However, various identi􀀂er spoo􀀂ng attacks [2],[3] have been exploited to deceive the ","cbCairrwxuSA0mXZ","https://ap.wps.com/l/cbCairrwxuSA0mXZ","pdf",409000,1,"English","en",105,"# Abstract\n# Introduction\n## IoT device identification needs\n## Limitations of identifier-based approaches\n## Progress and challenges of ML-based fingerprinting\n## Security gap and motivation for adversarial attack","[{\"question\":\"What is IoTGAN’s goal in machine learning based IoT device identification?\",\"answer\":\"IoTGAN aims to manipulate an IoT device’s traffic so the identification system evades machine learning based device identification. The attack targets the traffic features the model relies on.\"},{\"question\":\"How does IoTGAN handle the black-box setting when learning the discriminative model?\",\"answer\":\"IoTGAN uses a neural network based substitute model to fit the target model in black-box settings. The substitute model provides the discriminative capability needed for the attack.\"},{\"question\":\"How does IoTGAN evade identification without affecting IoT device functionality?\",\"answer\":\"A manipulative model is trained to add adversarial perturbations into IoT traffic through the substitute model. The perturbations are designed to evade identification while not influencing the device’s functionality.\"}]","IoTGAN - GAN Powered Camouflage Against Machine Learning Based IoT Device Identification | PDF",1785733119,20,{"code":4,"msg":30,"data":31},"ok",{"site_id":23,"language":22,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":85,"head_meta":87,"extra_data":89,"updated_unix":27},"iotgan-gan-powered-camouflage-against-machine-learning-based-iot-device-identification","",{"@graph":35,"@context":84},[36,53,67],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/iotgan-gan-powered-camouflage-against-machine-learning-based-iot-device-identification/120971/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":22,"description":14,"dateModified":61,"datePublished":61,"encodingFormat":60,"isAccessibleForFree":62,"interactionStatistic":63},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-08-03",true,{"@type":64,"interactionType":65,"userInteractionCount":20},"InteractionCounter",{"@type":66},"ViewAction",{"@type":68,"mainEntity":69},"FAQPage",[70,76,80],{"name":71,"@type":72,"acceptedAnswer":73},"What is IoTGAN’s goal in machine learning based IoT device identification?","Question",{"text":74,"@type":75},"IoTGAN aims to manipulate an IoT device’s traffic so the identification system evades machine learning based device identification. The attack targets the traffic features the model relies on.","Answer",{"name":77,"@type":72,"acceptedAnswer":78},"How does IoTGAN handle the black-box setting when learning the discriminative model?",{"text":79,"@type":75},"IoTGAN uses a neural network based substitute model to fit the target model in black-box settings. The substitute model provides the discriminative capability needed for the attack.",{"name":81,"@type":72,"acceptedAnswer":82},"How does IoTGAN evade identification without affecting IoT device functionality?",{"text":83,"@type":75},"A manipulative model is trained to add adversarial perturbations into IoT traffic through the substitute model. The perturbations are designed to evade identification while not influencing the device’s functionality.","https://schema.org",{"og:url":51,"og:type":86,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":88,"canonical":51},"index,follow",{"doc_id":7,"site_id":23},{"code":4,"msg":5,"data":91},[92,96,100,104,109,114,119,122,126,129,133],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":93,"show_sort_weight":94,"slug":95},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":97,"show_sort_weight":98,"slug":99},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":101,"show_sort_weight":102,"slug":103},"Exam",70,"exam",{"id":105,"doc_module":4,"doc_module_name":45,"category_name":106,"show_sort_weight":107,"slug":108},5,"Comic",60,"comic",{"id":110,"doc_module":4,"doc_module_name":45,"category_name":111,"show_sort_weight":112,"slug":113},6,"Technology",50,"technology",{"id":115,"doc_module":4,"doc_module_name":45,"category_name":116,"show_sort_weight":117,"slug":118},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":120,"slug":121},30,"research-report",{"id":123,"doc_module":4,"doc_module_name":45,"category_name":124,"show_sort_weight":28,"slug":125},9,"Religion & Spirituality","religion-spirituality",{"id":28,"doc_module":4,"doc_module_name":45,"category_name":127,"show_sort_weight":28,"slug":128},"World Cup","world-cup",{"id":130,"doc_module":4,"doc_module_name":45,"category_name":131,"show_sort_weight":130,"slug":132},10,"Lifestyle","lifestyle",{"id":134,"doc_module":4,"doc_module_name":45,"category_name":135,"show_sort_weight":105,"slug":136},19,"General","general"]