Jun Zhang, Jue Wang, Huan Li, Lidan Shou, Ke Chen, Gang Chen, Sharad Mehrotra. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.

In this paper, we address the role of middleware in enabling robust and resilient cyber-physical systems (CPSs) of the future. In particular, we will focus on how adaptation services can be used to improve dependability in instrumented cyber-physical systems based on the principles of “computational reflection.” CPS environments incorporate a variety of sensing and actuation devices in a distributed architecture; such a deployment is used to create a digital representation of the evolving physical world and its processes for use by a broad range of applications. CPS applications, in particular, mission critical tasks, must execute dependably despite disruptions caused by failures and limitations in sensing, communications, and computation. This paper discusses a range of applications, their reliability needs, and potential dependability holes that can cause performance degradation and application failures. In particular, we distinguish between the notion of infrastructure and information dependability and illustrate the need to formally model and reason about a range of CPS applications and their dependability needs. Formal methods based tools can help us design meaningful cross-layer adaptation techniques at different system layers of the CPS environment and thereby achieve end-to-end dependability at both the infrastructure and information levels.

Data outsourcing allows data owners to keep their data at untrusted clouds that do not ensure the privacy of data and/or computations. One useful framework for fault-tolerant data processing in a distributed fashion is MapReduce, which was developed for trusted private clouds. This paper presents algorithms for data outsourcing based on Shamir's secret-sharing scheme and for executing privacy-preserving SQL queries such as count, selection including range selection, projection, and join while using MapReduce as an underlying programming model. Our proposed algorithms prevent an adversary from knowing the database or the query while also preventing output-size and access-pattern attacks. Interestingly, our algorithms do not involve the database owner, which only creates and distributes secret-shares once, in answering any query, and hence, the database owner also cannot learn the query. Logically and experimentally, we evaluate the efficiency of the algorithms on the following parameters: (i) the number of communication rounds (between a user and a server), (ii) the total amount of bit flow (between a user and a server), and (iii) the computational load at the user and the server.