P. Kumari, B. M. S. B. Talukder, S. Sakib, B. Ray, Md. Tauhidur Rahman
{"title":"Independent detection of recycled flash memory: Challenges and solutions","authors":"P. Kumari, B. M. S. B. Talukder, S. Sakib, B. Ray, Md. Tauhidur Rahman","doi":"10.1109/HST.2018.8383895","DOIUrl":null,"url":null,"abstract":"Counterfeiting electronic components is a serious problem for the security and reliability of any electronic systems. Use of counterfeit or reused components not only impacts profit but also has a detrimental impact on several critical applications including aerospace, medical, and defense. To worsen the situation the number of counterfeiting components has increased considerably after the introduction of horizontal semiconductor supply chain. In this paper, we will focus on detecting recycled Flash memory, a major target of the counterfeiters because of its presence in the most electronic systems. Failure of the Flash memory in critical applications can have catastrophic effects. Detection of recycled Flash with high confidence i s challenging due to the variability among the different Flash chips caused by process variations. There is very few work on detecting recycled memory chips, and unfortunately, all of them require an extensive database to maintain which is impossible for several electronic systems. In this paper, we propose a new method for detecting fake Flash memory without the need for any prior database. Our method is based on statistical distribution of various Flash timing characteristics such as erase, program and read time on a fresh Flash IC. It has been found that timing characteristics are highly sensitive to memory usage (typically quantified as the program-erase count of a memory block) compared to the process variations. We demonstrate our method by characterizing the block to block timing variation on commercial off the shelf Flash ICs and compared it with the recycled or used one. Our method can identify a recycled IC of minimal usage (∼3.0%) with nearly 100% accuracy without requiring any prior database.","PeriodicalId":6574,"journal":{"name":"2018 IEEE International Symposium on Hardware Oriented Security and Trust (HOST)","volume":"34 1","pages":"89-95"},"PeriodicalIF":0.0000,"publicationDate":"2018-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"14","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2018 IEEE International Symposium on Hardware Oriented Security and Trust (HOST)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/HST.2018.8383895","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 14
Abstract
Counterfeiting electronic components is a serious problem for the security and reliability of any electronic systems. Use of counterfeit or reused components not only impacts profit but also has a detrimental impact on several critical applications including aerospace, medical, and defense. To worsen the situation the number of counterfeiting components has increased considerably after the introduction of horizontal semiconductor supply chain. In this paper, we will focus on detecting recycled Flash memory, a major target of the counterfeiters because of its presence in the most electronic systems. Failure of the Flash memory in critical applications can have catastrophic effects. Detection of recycled Flash with high confidence i s challenging due to the variability among the different Flash chips caused by process variations. There is very few work on detecting recycled memory chips, and unfortunately, all of them require an extensive database to maintain which is impossible for several electronic systems. In this paper, we propose a new method for detecting fake Flash memory without the need for any prior database. Our method is based on statistical distribution of various Flash timing characteristics such as erase, program and read time on a fresh Flash IC. It has been found that timing characteristics are highly sensitive to memory usage (typically quantified as the program-erase count of a memory block) compared to the process variations. We demonstrate our method by characterizing the block to block timing variation on commercial off the shelf Flash ICs and compared it with the recycled or used one. Our method can identify a recycled IC of minimal usage (∼3.0%) with nearly 100% accuracy without requiring any prior database.