Harrison Williams

Sagar Bharadwaj, Harrison Williams, Luke Wang et al.

World-scale augmented reality (AR) applications need a ubiquitous 6DoF localization backend to anchor content to the real world consistently across devices. Large organizations such as Google and Niantic are 3D scanning outdoor public spaces in order to build their own Visual Positioning Systems (VPS). These centralized VPS solutions fail to meet the needs of many future AR applications -- they do not cover private indoor spaces because of privacy concerns, regulations, and the labor bottleneck of updating and maintaining 3D scans. In this paper, we present OpenFLAME, a federated VPS backend that allows independent organizations to 3D scan and maintain a separate VPS service for their own spaces. This enables access control of indoor 3D scans, distributed maintenance of the VPS backend, and encourages larger coverage. Sharding of VPS services introduces several unique challenges -- coherency of localization results across spaces, quality control of VPS services, selection of the right VPS service for a location, and many others. We introduce the concept of federated image-based localization and provide reference solutions for managing and merging data across maps without sharing private data.

Harrison Williams, Alexander Lind, Kishankumar Parikh et al.

In order to service an ever-growing base of legacy electronics, both government and industry customers must turn to third-party brokers for components in short supply or discontinued by the original manufacturer. Sourcing equipment from a third party creates an opportunity for unscrupulous gray market suppliers to insert counterfeit devices: failed, knock-off, or otherwise inferior to the original product. This increases the supplier's profits at the expense of reduced performance/reliability of the customer's system. The most challenging class of counterfeit devices to detect is recycled counterfeits: recovered genuine devices which are re-sold as new. Such devices are difficult to detect because they typically pass performance and parametric tests but fail prematurely due to age-related wear. To address the challenge of detecting recycled devices pre-deployment, we develop Silicon Dating: a low-overhead classifier for detecting recycled integrated circuits using Static Random-Access Memory (SRAM) power-on states. Silicon Dating targets devices with no known-new record or purpose-built anti-recycling hardware. We observe that over time, software running on a device imprints its unique data patterns into SRAM through analog-domain changes; we measure the level and direction of this change through SRAM power-on state statistics. In contrast to highly symmetric power-on states produced by variation during SRAM fabrication, we show that embedded software data is generally highly asymmetric and that the degree of power-on state asymmetry imprinted by software reveals device use. Using empirical results from embedded benchmarks running on several microcontrollers, we show that Silicon Dating identifies recycled devices with 84.1% accuracy with no software-specific knowledge and with 92.0% accuracy by incorporating software knowledge---without prior device enrollment or modification.