A novel clustering technique using backscattering side channel for counterfeit IC detection

Over the past few years, globalization of the semiconductor supply chain has led companies to outsource much of the production cycle for integrated circuits (ICs). While outsourcing helps companies significantly reduce their cost and time-to-market, it also introduces concerns about the trustworthiness of an IC. One of the most serious problems is counterfeiting of ICs, which not only negatively impacts innovation and economic growth of the IC industry, but also creates serious threats and risks for systems that incorporate those counterfeit ICs. This paper proposes a novel method that uses the backscattering side-channel to cluster ICs such that counterfeits are separated from legitimate ICs. The backscattering side-channel, which has been introduced only recently, has been proven to outperform other side-channels in detecting hardware Trojan horses (HTs), i.e. ICs where additional logic gates (and connections to existing logic gates) have been added. In this work we use it to robustly separate ICs into legitimate and counterfeit ones, even when only layout or placement of the IC has changed, without any added logic or connections. We evalute our technique on a set of ten boards over six different counterfeit IC designs, and find that our technique tolerates manufacturing variations among different hardware instances, detecting counterfeit ICs with 100% accuracy and 0% false positives.