Progress 02/01/21 to 01/31/25
Outputs
Target Audience:The target application audience for our project is primarily system builders for CPS/IoT applications. In addition, a second target audience for our research has been the distributed computing, security, and dependability research community. By publishing our advances at conferences in these domains, the team has subjected our project work to be reviewed by expert peers and is exposing more researchers to the challenges faced in applications of CPS/IoT technology. Changes/Problems:
Nothing Reported
What opportunities for training and professional development has the project provided?This project served as a locus for the research activities of one postdoctoral researcher and three Ph.D. students. These researchers attended conferences to represent the research conducted in this project and delivered the official conference presentations on those advances. One of these students completed his Ph.D. based on his research conducted in this project and is now employed at Visa Research. The postdoctoral researcher who was supported by this project is now a professor at the University of Illinois at Chicago. This project also provided Prof. Reiter opportunities to interact with his Purdue colleagues (including visiting Purdue), and through these interactions he learned more about the needs of the agriculture community that the technologies pursued in this project might be leveraged to address. Concerns specific to CPS/IoT systems have driven much of the research in this project. Prof. Reiter looks forward to working with his Purdue colleagues to further this research and to expose it to stakeholders for whom it would be useful. How have the results been disseminated to communities of interest?All of our research has been disseminated through peer-reviewed publication at conferences in the areas of distributed computing or computer security and/or dependability, along with accompanying conference presentation. These papers have been uploaded to PubAg and are also available from publishers' digital libraries and from my web page (https://reitermk.github.io). Many of these papers are accompanied by reference implementations demonstrating the advances. What do you plan to do during the next reporting period to accomplish the goals?
Nothing Reported
Impacts
What was accomplished under these goals?
In the outcomes and impact statement below, the outcomes are of the "changes in knowledge" type. While other researchers are building upon our results, we are not aware of their adoption yet in practice (changes of action) or resulting changes in condition. Below, each label (a, b, ...) refers to the same contribution and roughly corresponds to one publication uploaded to PubAg. We omitted the "Data collected" descriptions below, since we did not collect field data on which we evaluated our research. 1. Fault Tolerant Analytics. - Major activities completed: a) We developed Byzantine fault-tolerant (BFT) agreement protocols for partially synchronous networks that tolerate fewer than one-half Byzantine faults without worsening communication complexity, by leveraging small trusted hardware at each node. b) We proposed a "Unique Chain Rule" (UCR) as a primitive for building efficient BFT agreement protocols (without using trusted hardware). This is a simple agreement rule where extending a block by including its hash in the next block, is treated as a vote for the proposed block and its ancestors. c) We initiated study of the energy efficiency of BFT agreement protocols and proposed a novel protocol that optimizes energy for the common case. d) We developed HashRand, an efficient asynchronous random beacon protocol that generate beacons using only hash functions and pairwise channels. HashRand is efficient and post-quantum secure, owing to its use of hash computations (vs. discrete-log cryptography used by others). - Discussion of results: a) BFT agreement protocols for partially synchronous networks tolerate fewer than one-third Byzantine faults, and while small trusted hardware had been used in prior work to increase that limit to one-half, it did so at the cost of increased communication complexity. Our work showed that this increase is not necessary. b) Most existing BFT agreement protocols relied on constructs inherently requiring O(n^2) communication overhead among n nodes. This limitation is overcome by the UCR. c) Today, state-of-the-art agreement protocols are designed to reduce latency or communication overhead. Agreement protocols not optimized for energy efficiency reduce the lifetime of CPS, and our work initiated exploration of energy efficiency for agreement in CPS. d) Regular access to bias-resistant randomness is important for secure distributed computing. Distributed random beacon protocols address this need by distributing trust across multiple nodes, with the majority of them assumed honest. Existing beacon systems exhibit high computational costs or expect the network to be partially or fully synchronous, whereas our work requires neither. - Key outcomes: a) Our work provides efficient protocols that can leverage numerous types of commercially available, trustworthy add-ons (TPMs, YubiKeys, smartcards, etc.) that are simpler to implement than secure enclaves for arbitrary computation (e.g., Intel SGX). b) We developed three applications of UCR. The first two are two novel synchronous protocols with linear best-case communication. In the third, we used UCR to make any protocol's commits publicly verifiable, so clients no longer have to wait for 2f+1 confirmations on every block, where f is the allowed number of Byzantine faults. c) We presented an application of our energy-efficiency framework in CPS. We analytically determined the parameter ranges for when our proposed protocol offers better energy efficiency versus a baseline utilizing an external trusted node. Finally, we demonstrated our approach's practicality by analyzing our protocol's energy efficiency through experiments on a CPS test bed. d) In a geo-distributed testbed of n=136 nodes, HashRand produced beacons at a rate >5x higher than a state-of-the-art competitor. We also demonstrated the utility of HashRand by using it to build a post-quantum secure asynchronous agreement protocol. 2. Approximate Consensus: - Major activities completed: e) We developed SensorBFT, an energy-efficient, fault-tolerant approach for target localization. SensorBFT uses a novel asynchronous approximate agreement protocol that enables correct sensors to achieve an approximate consensus despite faulty sensors. Sensors fulfill their energy budgets by tuning the precision and accuracy of localization. f) When sensor nodes measure a common source, reaching an agreed-upon output within the convex range of correct inputs, known as convex validity, is imperative. We developed Delphi, a new protocol that leverages the fact that in CPS, the ambient noise causing the difference between inputs follows distributions like Normal or Lognormal. So, given that honest parties often measure inputs that are close to each other, we assume that honest inputs are sampled from a thin-tail distribution. - Discussion of results: e) Prior systems implementing target localization do not allow for sensors to exhibit Byzantine faults. While many algorithms can be naively used to implement target localization in a fault-tolerant manner, these approaches are energy-intensive and are not appropriate for CPS. SensorBFT addresses this problem. f) Existing approximate agreement protocols achieving convex validity either use common coins that themselves are computationally expensive to implement, or leverage asynchronous approximate agreement, for which known solutions have high communication costs. Delphi sidesteps these limitations with a different approach. - Key outcomes: e) In good scenarios, SensorBFT reduces communication from O(n^3) to O(n^2) messages per round, where n is the number of sensors. In a sensor testbed with n=19 sensors, SensorBFT consumes 2/5th the energy consumed by existing solutions for only a 2% loss in accuracy. f) Delphi is a deterministic protocol with ~O(n^2) communication and minimal computation overhead. Delphi assumes that honest inputs are bounded, except with negligible probability. For an n=160-node system, Delphi achieves an 8x and 3x improvement in latency within CPS and AWS environments, respectively, relative to baseline protocols. 3. Privacy-preserving Distributed Computing: - Major activities completed: g) We initiated the study of cryptographic protocols that enable two parties observing regions (or objects) in an arbitrary-dimension Euclidean space to privately detect if the regions overlap and approximate the volume of the overlapping region. h) We conducted the first critical analysis of building secure, performant, and private BFT consensus by integrating off-the-shelf crash fault-tolerant (CFT) protocols with trusted execution environments (TEEs). - Discussion of results: g) Existing tools do not address how to privately detect region overlaps or degree of overlap. They either detect only exact matches of elements or are designed for point comparisons, not regions. h) We showed that naively implementing a CFT protocol inside a TEE (Intel SGX) does not achieve private BFT agreement. - Key outcomes: g) We benchmarked our protocol on data generated from the CARLA autonomous driving simulator and the ScanNet 3D image dataset. Our protocol outperforms baselines and, e.g., takes ~0.5s to estimate the volume of overlap of two 3D boxes. h) We developed Engraft, a secure TEE-guarded CFT implementation. It achieves consensus if fewer than half of the nodes exhibit Byzantine faults (while their enclaves do not). It also supports confidential server state, in support of private distributed computing. Impact Statement: These works have opened new possibilities to achieve performant BFT CPS, including ones that support private state. Despite being recent, they have been cited >85 times, indicating that they are inspiring further research, as well.
Publications
|
Progress 02/01/23 to 01/31/24
Outputs
Target Audience:The target application audience for our project is primarily system builders for CPS/IoT applications. In addition, a second target audience for our research has been the distributed computing, security, and dependability research community. By publishing our advances at conferences in these domains, the team has subjected our project work to be reviewed by expert peers and is exposing more researchers to the challenges faced in applications of CPS/IoT technology. Changes/Problems:
Nothing Reported
What opportunities for training and professional development has the project provided?Research related to this grant has served as the locus of research for two Ph.D. students this year. How have the results been disseminated to communities of interest?Dissemination has been through the peer-reviewed publications listed under "Products". In each case, the publication was accompanied by a presentation to the conference audience, as well. What do you plan to do during the next reporting period to accomplish the goals?We are continuing our efforts onprivacy-preserving distributed computing. One project we are continuing to develop is enforcing policies on the data that a party contributes to a distributed computation without disclosing those data to others. Another direction we are developing is how to apply advanced formal methods to verify the correctness of distributed protocols, with particular attention to some of the protocols we have already developed in this effort.
Impacts
What was accomplished under these goals?
Agreement protocols are crucial in various emerging applications, spanning from distributed (blockchains) oracles to fault-tolerant cyber-physical systems. In scenarios where sensor nodes measure a common source, maintaining output within the convex range of correct inputs, known as convex validity, is imperative. Present asynchronous convex agreement protocols employ either randomization, incurring substantial computation overhead, or approximate agreement techniques, leading to high communication overhead. We developed Delphi, a deterministic protocol with lower communication and minimal computation overhead. Delphi assumes that honest inputs are bounded, except with negligible probability, and integrates agreement primitives from the literature with a novel weighted averaging technique. We demonstrated Delphi's superior performance, showcasing a significantly lower latency compared to state-of-the-art protocols. For example, for a 160-node system, Delphi achieves an 8x and 3x improvement in latency within CPS and AWS environments, respectively. We also explored the energy efficiency of agreement protocols, specifically for Byzantine fault-tolerant State Machine Replication (SMR). We proposeda novel SMR protocol that optimizes energy efficiencey for the steady state, i.e., when the leader is correct. This is done by reducing the number of required signatures per consensus unit and the communication complexity by order of the number of nodes,compared to the state-of-the-art BFT-SMR solutions. Second, we presented an application in the cyber-physical system (CPS) setting, where we consider a partially connected system by optionally leveraging wireless multicasts among neighbors. We analytically determined the parameter ranges where our proposed protocol offers better energy efficiency than communicating with a baseline protocol utilizing an external trusted node. We presented a hypergraph-based network model and generalized previous fault-tolerance results to the model. Finally, we demonstrated our approach's practicality by analyzing our protocol's energy efficiency through experiments on a CPS test bed. In particular, we observed as high as 64% energy savings when compared to the state-of-the-art SMR solution for 10 nodes using BLE. Regular access to unpredictable and bias-resistant randomness is important for numerous types of distributed agreement. Distributed random beacon protocols address this need by distributing trust across multiple nodes, with the majority of them assumed to be honest. Many current random beacon systems rely on threshold cryptographic setups or exhibit high computational costs, while others expect the network to be partially synchronous or synchronous. To overcome these limitations, we developed HashRand, a computation- and communication-efficient asynchronous random beacon protocol that requires only secure hashing and pairwise secure channels to generate beacons. The computational efficiency of HashRand is attributed to the two-orders-of-magnitude lower cost of a one-way hash computation compared to discrete-log exponentiation. Interestingly, besides reduced overhead, HashRand achieves post-quantum security by leveraging a hash function secure against quantum adversaries, setting it apart from other random beacon protocols that use discrete-log cryptography. In a geo-distributed testbed of 136 nodes, HashRand produces 78 beacons per minute, which is at least 5x higher than a recent competitor. We also demonstrated the practical utility of HashRand by implementing a post-quantum secure asynchronous SMR protocol using it. Finally, target localization is used for detecting and locating an adverse event called a target in a geographic area. This primitive is applicable in the physical security domain (e.g., detecting intruders in an area) or for disaster preemption, such as detecting ignition events of forest fires. Prior systems implemented this primitive over large areas by deploying a network of sensor devices, which detect changes in a specific physical parameter like pressure or temperature induced by a target. However, these systems are not designed for use in adverse environments where one or more sensors can behave in a faulty manner. While many algorithms in the distributed systems literature can be naively used to implement target localization in a fault-tolerant manner, these approaches are energy-intensive as they use computationally expensive cryptographic operations not appropriate for resource-constrained sensors. We designed SensorBFT, an energy-efficient, fault tolerant approach for target localization. SensorBFT uses a novel asynchronous approximate agreement protocol that enables correct sensors to achieve an approximate consensus in the presence of faulty sensors. Sensors fulfill their energy budgets by tuning the precision and accuracy of localization, where precision is the difference between honest sensors' outputs and accuracy is the difference between an honest sensor's output and the target's true location. In a sensor testbed with 19 sensors, SensorBFT consumes 2/5th the energy consumed by existing solutions for a minor 2% loss in accuracy, significantly enhancing efficiency and coverage.
Publications
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2023
Citation:
A. Bhat, A. Bandarupalli, M. Nagaraj, S. Bagchi, A. Kate, and M. K. Reiter. EESMR: Energy efficient BFT-SMR for the masses. In Proceedings of the 24th ACM/IFIP International Middleware Conference, December 2023.
- Type:
Conference Papers and Presentations
Status:
Accepted
Year Published:
2024
Citation:
A. Bandarupalli, A. Bhat, S. Bagchi, A. Kate, C.-D. Liu-Zhang, and M. K. Reiter. Delphi: Efficient asynchronous approximate agreement for distributed oracles. In Proceedings of the 54th IEEE/IFIP International Conference on Dependable Systems and Networks, June 2024.
- Type:
Conference Papers and Presentations
Status:
Accepted
Year Published:
2024
Citation:
A. Bandarupalli, A. Bhat, S. Chaterji, M. K. Reiter, A. Kate and S. Bagchi. SensorBFT: Fault-tolerant target localization using Voronoi diagrams and approximate agreement. In Proceedings of the 44th IEEE International Conference on Distributed Computing Systems, July 2024.
- Type:
Conference Papers and Presentations
Status:
Submitted
Year Published:
2024
Citation:
A. Bandarupalli, A. Bhat, S. Bagchi, A. Kate, and M. K. Reiter. Random beacons in Monte Carlo: Efficient asynchronous random beacon without threshold cryptography. In Proceedings of the 31st ACM Conference on Computer and Communications Security, October 2024.
|
Progress 02/01/22 to 01/31/23
Outputs
Target Audience:The target application audience for our project is primarily system builders for CPS/IoT applications. In addition, a second target audience for our research has been the distributed computing, security, and dependability research community. By publishing our advances at conferences in these domains, the team has subjected our project work to be reviewed by expert peers and is exposing more researchers to the challenges faced in applications of CPS/IoT technology. Changes/Problems:
Nothing Reported
What opportunities for training and professional development has the project provided?The research described above on privacy-preserving distributed computing was the focus of a postdoctoral researcher at Duke, working with Prof. Reiter. This project has brought him into the domain of IoT technologies and enabled him to apply his cryptographic expertise there. The research on the Unique Chain Rule was performed in conjunction with a graduate student at Purdue University, with whom Prof. Reiter collaborates extensively. The research on leveraging TEEs to achieve BFT from CFT protocols was performed in conjuction with a student who Prof. Reiter intends to admit to the Ph.D. program at Duke. How have the results been disseminated to communities of interest?Dissemination has primarily been through the peer-reviewed publications listed below. In each case, the publication was accompanied by a presentation to the conference audience. In addition, the research isexposed to application-domain expertsat Purdue University by research collaborators there. What do you plan to do during the next reporting period to accomplish the goals?In the next reporting period, we intend to broaden our study of supporting privacy-preserving distributed computing. One project we specifically intend to explore is enforcing policies on the data that a party contributes to a distributed computation without disclosing those data to others. Another direction we plan to explore is methods for implementing unpredictable and bias-resistant randomness, which is important for secure distributed computing. Distributed random beacon protocols address this need by distributing trust for randomness generation across multiple nodes, with the majority of them assumed to be honest. However, current random beacon systems suffer from a numberof drawbacks, which we will seek to rectify.
Impacts
What was accomplished under these goals?
We developed the first critical analysis of building highly secure, performant, and confidential Byzantine fault-tolerant (BFT) state-machine replication (SMR) by integrating off-the-shelf crash fault-tolerant (CFT) protocols with trusted execution environments (TEEs). TEEs, like Intel SGX, are CPU extensions that offer applications a secure execution environment with strong integrity and confidentiality guarantees, by leveraging techniques like hardware-assisted isolation, memory encryption, and remote attestation. It has been speculated that when implementing a CFT protocol inside Intel SGX, one would achieve security properties similar to BFT. However, we showed in this work that simply combining CFT with SGX does not directly yield a secure SMR protocol, given the wide range of attack vectors on SGX. We systematically studied the fallacies in such a strawman design by performing model checking, and proposed solutions to enforce safety and liveness. We also developedEngraft, a secure enclave-guarded Raft implementation that, firstly, achieves SMR on a cluster of 2f+1 machines tolerating up to f nodes exhibiting Byzantine-faulty behavior (but well-behaved enclaves); secondly, offers confidentiality for privacy-preserving SMR; and finally, allows the reuse of a production-quality Raft implementation, BRaft, in the development of a highly performant BFT system. Most existing BFT SMR protocols not leveraging TEEs rely explicitly on either equivocation detection or quorum certificate formations to ensure protocol safety. These mechanisms inherently require O(n2) communication overhead among n participating servers. We proposed the Unique Chain Rule (UCR), a simple rule for hash chains where extending a block by including its hash in the next block, is treated as a vote for the proposed block and its ancestors. When a block obtains a vote from at least one correct server, we can commit the block and its ancestors. While this idea was used implicitly earlier in conjunction with equivocation detection or quorum certificate generation, this work employs it explicitly to show safety. We developed three applications of UCR. Two, named Apollo and Artemis, are novel synchronous SMR protocols with linear best-case communication complexity using round-robin and stable leaders, respectively. In the third, we employ UCR in a black-box fashion toward making any SMR commits publicly verifiable, where clients will no longer have to wait for 2f+1 confirmations on every block but can instead collect a UCR proof consisting of min(k,f)+1 extensions on a block, where k is a security parameter and f is the number of Byzantine faults tolerated by the protocol. This results in faster syncing times for clients as the publicly verifiable proofs can also be gossiped with every new block extension confirming a new block. Finally, we made progress during this reporting period on privacy-preserving distributed computing. Advances in computer vision have made it possible to accurately map objects as regions in 3-dimensional space using LIDAR point clouds. These systems are key building blocks of several emerging CPStechnologies. Comparing and validating the output of sensors at different vantage points observing the same scenery can enable these systems to detect faults, identify common obstacles, and improve decision making. However sharing sensor outputs among mutually untrusting parties can leak unwanted information, e.g., model parameters or relative location of the sensors. We initiated the study of cryptographic protocols that enable two parties observing regions (or objects) in an arbitrary-dimension Euclidean space to privately detect if the regions overlap and approximate the volume of the overlapping region. The protocols rely only on cheap symmetric-key primitives and feature reasonable communication costs and compute times. We benchmarked implementations of these protocols and showed, e.g., that it takes roughly 0.5 seconds to approximate the volume of the overlapping region of two 3D boxes with low error probability.
Publications
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2022
Citation:
W. Wang, S. Deng, J. Niu, M. K. Reiter, and Y. Zhang. ENGRAFT: Enclave-guarded Raft on Byzantine faulty nodes. In Proceedings of the 29th ACM Conference on Computer and Communications Security, November 2022.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2022
Citation:
S. Yandamuri, I. Abraham, K. Nayak, and M. K. Reiter. Communication-efficient BFT using small trusted hardware to tolerate minority corruption. In Proceedings of the 26th International Conference on Principles of Distributed Systems, December 2022.
- Type:
Conference Papers and Presentations
Status:
Accepted
Year Published:
2023
Citation:
A. Bhat, A. Bandarupalli, S. Bagchi, A. Kate, and M. K. Reiter. The unique chain rule and its applications. In Proceedings of the 27th International Conference on Financial Cryptography and Data Security, May 2023.
- Type:
Conference Papers and Presentations
Status:
Submitted
Year Published:
2023
Citation:
A. Chakraborti and M. K. Reiter. Privately evaluating region overlaps with applications to collaborative sensor output validation. Submitted to the 8th IEEE European Symposium on Security and Privacy, July 2023.
|
Progress 02/01/21 to 01/31/22
Outputs
Target Audience:The target application audience for our project is primarily system builders for CPS/IoT applications that could improve farm efficiency. Through the Wabash Heartland Innovation Network (WHIN) project, PD Chaterji is engaged with several farmers, putting wireless mesh network sensor systems and gateway nodes in their farms, in Tippecanoe, Benton, and White counties in Indiana. Our outreach involves recorded videos showing the deployments of rugged, packaged sensor nodes and gateway nodes (NVIDIA Jetsons and Raspberry Pis) in the WHIN 10-county region, in Purdue's experimental farms (830 managed acres of Throckmorton Purdue Agricultural Center (TPAC) and 1134 acres of Agronomy Center for Research and Education (ACRE)), and the Birck Nanotechnology Center; the latter affords the sensor and gateway nodes, a laboratory environment for testing and certification. The deployments in the farms have been in partnership with the farmers and currently use the farmers' WiFi for connectivity from the gateway node to the hubs. In addition to this application domain, a second target audience for our research has been the distributed computing, security, and dependability research community. By publishing our advances at conferences in these domains, the team has subjected our project work to be reviewed by expert peers and is exposing more researchers to the challenges faced in applications of CPS/IoT technology. Changes/Problems:
Nothing Reported
What opportunities for training and professional development has the project provided?Adithya Bhat, Akhil Sai Bandarupalli, Jayoung Lee, Chen-Yi Lu, Easwar V Mangipudi, and PengCheng Wang are six Ph.D. students involved in this project at Purdue University. Sravya Yandamuri is a student supported by this project at Duke University. Each of these students communicates with two or more PDs through weekly video calls. Our students have attended conferences (e.g.,CCS 2021, SOSP 2021) to present their papers as well as general attendees; however, due to the ongoing pandemic, they have attended the conferences virtually. PD Kate hosted an undergraduate research intern (Karol Stephany Insuasty Mejia) from Colombia (South America) for four months (August-December 2021). The student studied the implementation of homomorphic encryption techniques in hardware toward developing energy-efficient privacy-preserving solutions. This is relevant to our third thrust. Working with PD Chaterji, high-schooler Sarthak Jain and sophomore Akash Melachuri are developing a frontend rendering counterpart for the project's computer vision backend algorithms for users to enter the service level objectives, which could be farmers or AgTech clients entering the requirements to adhere to in terms of accuracy, latency, energy bounds under which our algorithms will execute, example presentation during Covid over Zoom here: https://schaterji.io/events/frontend.html John Scott, an Extension drone engineer, is a part of Prof. Chaterji's lab's (Innovatory for Cells and Neural Machines, ICAN's) efforts, along with her graduate student Chen-Yi Lu, to use her technology on farms and her students with John Scott have been deploying drone flights using her automated drone descent technology. Here, the drone optimizes the height of the flight based on the coverage and precision required for scouting the agricultural fields. This is being done on Purdue's farms at ACRE: https://ag.purdue.edu/agry/acre/Pages/default.aspx How have the results been disseminated to communities of interest?Dissemination has been through the peer-reviewed publications listed below. Also, we have engaged with Extension specialists and Lilly Endowment's Wabash Heartland Innovation Network (WHIN) to demonstrate our technology to AgTech (Bayer Crop Sciences) and to farmers in the area. PD Chaterji also writes blog posts for dissemination of her computer vision on IoT devices and drone technology to a wider audience: #1: https://schaterji.io/blog/IoT-digag.html; #2: https://schaterji.io/blog/streaming-iot.html What do you plan to do during the next reporting period to accomplish the goals?Adithya Bhat and Akhil Sai Bandarupalli are developing energy-efficient consensus protocols suitable to be deployed in agriculture sensor networks. Akhil Sai Bandarupalli is also developing approximate consensus ideas towards efficiently collecting data from a sensor network. In particular, he is defining a new metric called convergence fairness to evaluate Byzantine Agreement (BA) protocols and designs an approximate agreement protocol that achieves non-zero convergence fairness without shared randomness in an asynchronous network. Sravya Yandamuri is currently working on an information theoretic leader election protocol for an asynchronous network that has sub quadratic communication complexity and equal work. A protocol that works in the asynchronous setting enables progress at the speed of the network and maintains safety regardless of any message delays. The purpose of obtaining sub quadratic communication complexity and equal work is scalability. Gathering farm scouting data through a combination of our computer vision technology and drone surveillance using smart algorithms for low-energy, high-coverage flights. Developing semi-supervised algorithms for learning patterns from farm data, especially leveraging the volumes of unlabeled farm data, with a focus on semantic segmentation useful for monitoring crop health. Collaboration with Microsoft Azure for IoT edge and cloud computing innovation for streamlining the collection of large volumes of farm data with suitable encryption techniques. PDs hold regular weekly meetings every week to discuss research findings and the team will continue to meet going forward.
Impacts
What was accomplished under these goals?
1. Fault Tolerant Analytics: First, the team is developing energy-optimized Byzantine-fault tolerant (BFT) state-machine replication (SMR) protocols for CPS nodes. The SMR abstraction, which is also the basis of all blockchains and cryptocurrencies, is generic enough to allow the execution of any computation task on the collected data. Our distributed consensus protocols resist malicious actors with a well-defined upper bound on the number of adversaries of different strengths. Second, the team is exploring the use of small trusted hardware primitives to improve the fault tolerance of BFT protocols to one-half faults, without increasing communication complexity. Our results include a version of the popular HotStuff SMR protocol that retains linear communication complexity in each view and a version of the VABA protocol with quadratic communication, both leveraging trusted hardware to tolerate a minority of corruptions. As a building block, we developed a communication-efficient provable broadcast, a core broadcast primitive with increased fault tolerance. 2. Approximate Consensus over Dynamic Networks: For dynamic CPS networks such as livestock-mounted CPS nodes, our goal will be to understand the fundamental connectivity lower-bounds first and then offer necessary and sufficient conditions for coordination and desired analytics. We have developed energy-optimized protocols to achieve approximate consensus over static networks. 3. Privacy-preserving Distributed Computing: The team is also developing privacy-aware distributed computation protocols among a distributed set of CPS nodes. We have observed some key computation and communication inefficient tasks/components from the existing secure multi-party computation (MPC) framework, and working towards reducing the overheads by making MPC suitable for the CPS networks.
Publications
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2021
Citation:
Sravya Yandamuri, Ittai Abraham, Kartik Nayak, and Michael Reiter. Brief announcement: Communication-efficient BFT using small trusted hardware to tolerate minority corruption. In Proceedings of the 35th International Symposium on Distributed Computing, October 2021.
- Type:
Other
Status:
Published
Year Published:
2021
Citation:
Adithya Bhat, Akhil Bandarupalli, Saurabh Bagchi, Aniket Kate, and Michael Reiter. Apollo -- Optimistically linear and responsive SMR. Cryptology ePrint Archive, no. 2021/180, 2021.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2022
Citation:
Thomas Yurek, Licheng Luo, Jaiden Fairoze, Aniket Kate, and Andrew Miller.
hbACSS: How to Robustly Share Many Secrets. In Network and Distributed System Security Symposium (NDSS), 2022.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2021
Citation:
Adithya Bhat, Nibesh Shrestha, Zhongtang Luo, Aniket Kate, and Kartik Nayak. RandPiper - Reconfiguration-Friendly Random Beacons with Quadratic Communication. In 28th ACM Conference on Computer and Communications Security (CCS), 2021.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2021
Citation:
Shikhar Suryavansh, Abu Benna, Chris Guest, and Somali Chaterji. Ambrosia -- Reduction in Data Transfer from Sensor to Server for Increased Lifetime of IoT Sensor Nodes. In Design Automation Conference (DAC) Work-in-Progress Poster, 2021. Nature Scientific Reports, 2021.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2022
Citation:
Ran Xu, Jayoung Lee, Pengcheng Wang, Saurabh Bagchi, Yin Li, and Somali Chaterji. LiteReconfig: Cost and content aware reconfiguration of video object detection systems for mobile GPUs. In ACM EuroSys, 2022.
- Type:
Other
Status:
Published
Year Published:
2021
Citation:
Pengcheng Wang, Edgardo Barsallo Yi, Tomas Ratkus, and Somali Chaterji.
ORPHEUS: Living Labs for End-to-End Data Infrastructures for Digital Agriculture. arXiv:2111.09422, October 2021.
|
|