Use Case

Multi-Sourced Data Analytics to Improve Food Production & Security

K. McCurdy (Sedgwick Reserve, University of California, Santa Barbara Natural Reserve System) with B. Roberts, B. Sethuramasamyraja (California State University, Fresno), B. Liu (California State Polytechnic University, San Luis Obispo), C. Krintz, R. Wolski (UC Santa Barbara)

What?

• Interdisciplinary analysis of data from a wide variety of sources, including the general public, to study and improve ecological outcomes

Why Aristotle?

• Scalable and portable infrastructure
• Access to multiple data sources, many of which are already in public and private clouds

Accomplishments

• Completed large-scale Google TensorFlow training and parallel image classification runs on Aristotle for the "Where's the Bear?" camera trap application analysis project (20 training runs using 2000 cores hours per run and a classification run using 1800 core hours to classify a test sample of 10,000 images prior to classifying 240,000 images from a single camera).
• Used soil moisture sensing devices, edge computing, and Aristotle to schedule vineyard irrigation (saved 66% of the water used previously).
• Installed hardware and software to instrument almond trees in a Fresno, CA test orchard to see how much water can be saved by irrigating the different sides of the root stock in proportion to its dryness.
• Developed innovative analytics with a minimally invasive instrumentation footprint for an Exeter, CA citrus test orchard to get highly accurate temperature readings at night when frost could form.
• Progress was made developing a low-power, low-cost, multi-tier IoT deployment for citrus frost prevention and the differential irrigation of almond trees.
• The Sedgwick Reserve "Where’s the Bear" project is using the Pacific Research Platform’s Kubernetes and containers environment to train image recognition models in conjunction with Aristotle.
• A new use case called Citrus Under Protective Screening (CUPS) was initiated. CUPS is a potential remedy for citrus greening which has devastated citrus in FL and is now threatening CA. Aristotle will serve the data hosting service for CUPS.
• 8 new publications (including a Best Student Paper award) and 2 keynote presentations were produced in PY4.
• Installed cameras and weather sensors at the UCSB Edible Campus farm site and developed the data acquisition system hosted on Aristotle.
• Installed sensors inside and outside the Citrus Under Protective Screening facility and deployed an IoT system that provides both real-time measurements via a visualization tool developed by Aristotle REU student Kareme Celik and a prototype frost alerting system.
• Developed the Aristotle AWS Pricing Tool to help users compare Aristotle resources to various AWS alternatives based on performance, cost, and price-performance.
• Developed 3 new collaborations that wish to leverage Aristotle infrastructure, tools and techniques: (1) Woods Hole Oceanographic Institute and UCSB Marine Sciences plan to study forest dynamics by fusing and analyzing remote sensing data from the seabed and atmosphere off the coast of CA, (2) DOE researchers are building ground-based cloud observation networks to study cloud formation and lifecycle dynamics and will use Aristotle artifacts to deploy two new observation networks at Sedgwick and Long Island, (3) the Math Dept. at Cal State Long Beach will use Aristotle science team expertise in using, maintaining, and scaling clouds to automate and scale its calculus curriculum.
• A new UCSB grant from the NSF titled "Detroit - A New End-to-End System for Practical and Accessible IoT" will leverage the Aristotle legacy.

Plans

• Develop a hybrid machine-learning and CFD model to support Citrus Under Protective Screening (CUPS) spraying operations and frost prevention.
• Enhance REU student Kerem Celik's telemetry data visualizer which works either as a local tool or as a service hosted in Aristotle.
• Develop sustainable land use practices at Sedgwick that employ livestock as part of the management lifecycle and install new monitoring infrastructure.
• Provide the instrumentation and analytics necessary to evaluate the first CUPS installation (at scale) in California. Aristotle will be providing the computational infrastructure to the team that is necessary to analyze the effects of CUPS on citrus production.

Products

.
• Zhang, M., Krintz, C. & Wolski, R. (2021). Sparta: A Heat-Budget-based Scheduling Framework on IoT Edge Systems. International Conference on Edge Computing.
• Zhang, M, Krintz, C. & Wolski, R. (2020). Edge-adaptable serverless acceleration for machine learning Internet of Things applications. Journal of Software: Practice and Experience..
• George, G., Bakir, F., Wolski, R. & Krintz, C. (2020). NanoLambda: Implementing Functions as a Service at all resource scales for the Internet of Things. P2020 IEEE/ACM Symposium on Edge Computing (SEC)..
• Dimopoulos, S., Krintz, C. & Wolski, R. (2020). Fair scheduling for deadline-driven, resource-constrained, multi-analytics workloads. 2020 International Conference on Computing, Networking and Communications (ICNC).
• Vaillancourt, P., Wineholt, B., Barker, B., Deliyannis, P., Zheng, J., Suresh, A., Brazier, A., Knepper, R. & Wolski, R. (2020). Reproducible and portable workflows for scientific computing and HPC in the cloud. Practice and Experience in Advanced Research Computing (PEARC'20).
• Wolski, R., Lifka, D. & Furlani, T. (2020). Data analysis and management for multi-campus cyberinfrastructure through cloud federation. Poster presentation at the 2020 NSF Cyberinfrastructure for Sustained Scientific Innovation (CSSI) Principal Investigator Meeting, Seattle, WA.
• Wolski, R., Krintz, C., Bakir, F., Lin, W.T. & George, G. (2019). Rethinking scalable services for the Internet of Things. Presentation at the Big Data and Extreme-Scale Computing Workshop (BDEC2), San Diego, CA.
• Wolski, R., Krintz, C., Bakir, F., George, G. & Lin, W.T. (2019). CSPOT: portable, multi-scale functions-as-a-service for IoT. Proceedings of the 4th ACM/IEEE Symposium on Edge Computing (SEC'19).
• Dimopoulos, S., Krintz, C. & Wolski, R. (2019). Towards distributed, fair, deadline-driven resource allocation for cloudlets. Proceedings of the 4th Workshop for Edge Clouds and Cloudlets.
• Golubovic, N., Krintz, C. & Wolski, R. (2019). A scalable system for executing and scoring K-means clustering techniques and its impact on applications in agriculture. International Journal of Big Data Intelligence.
• Bakir, F., Woksi, R., Krintz, C. & Ramachandran, G. (2019). Devices-as-services: Rethinking scalable service architectures for the internet of things. Proceedings of the 2019 USENIX Conference.
• Golubovic, N., Wolski, R., Krintz, C. & Mock, M. (2019). Improving the accuracy of outdoor temperature prediction by IoT devices. IEEE Conference on IoT.
• Zhang, M., Krintz, C. Wolski, R. & Mock, M. (2019). Seneca: Fast and low cost hyperparameter search for machine learning methods. IEEE Cloud 2019.
• Carson, K., Thomason, J., Wolski, R., Krintz, C. & Mock, M. (2019). Mandrake: Implementing durability for edge clouds. IEEE International Conference on Edge computing.
• Lin, W-t., Bakir, F., Krintz, C. & Mock, M. (2019). Data repair for distributed event-based IoT applications. Proceedings of the 13th ACM International Conference on Distributed and Event-based Systems.
• George, G., Wolski, R., Krintz, C. & Brevik, J. (2019). Analyzing AWS spot instance pricing (2019). IEEE International Conference on Cloud Engineering (IC2E).
• Krintz, C. SmartFarm: IoT systems that simplify and automate agriculture analytics. Keynote at 8th International Conference on Internet of Things (IoT 2018) Santa Barbara, CA.
• Krintz, C. Adventures and opportunities in cyber-physical systems and research. Keynote at 2018 International Conference on Computer Aided Design, San Diego, CA.
• Centaurus K-Means Clustering as a Service video (2018) demonstrates the balancing of a scalable clustering analysis workload between two Aristotle clouds.
• Lin, W.T., Krintz, C. & Wolski, R. (2018). Tracing function dependencies across clouds. Proceedings of the IEEE 11th International Conference on Cloud Computing.
• Jayathilaka, H., Krintz, C. & Wolski, R. (2018). Detecting performance anomalies in cloud platform applications. IEEE Transactions on Cloud Computing.
• Golubovic, N., Gill, A., Krintz, C. & Wolski, R. (2018). CENTAURUS: A cloud service for K-means clustering. 2017 IEEE 15 Intl Conf on Dependable, Autonomic and Secure Computing, 15th Intl Conf on Pervasive Intelligence and Computing, 3rd Intl Conf on Big Data Intelligence and Cyber Science and Technology
• Where’s the Bear article (2018) highlights how UCSB use case scientists Chandra Krintz & Rich Wolski are identifying wildlife with machine learning and their IoTDI publication details how this automated wildlife image processing uses IoT, edge cloud systems, and Aristotle.
• Wolski, R., Brevik, J., Chard, R. & Chard, K. (2017). Probabilistic guarantees of execution duration for Amazon spot instances. Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis – SC17.
• Dimopoulos, S., Krintz, C. & Wolski, R. (2017). PYTHIA: Admission control for multi-framework, deadline-driven, big data workloads. Proceedings of 2017 IEEE 10th International Conference on Cloud Computing – CLOUD.
• Wolski, R. & Brevik. J. (2017). QPRED: Using quantile predictions to improve power usage for private clouds. Proceedings of 2017 IEEE 10th International Conference on Cloud Computing – CLOUD.
• Wolski, R. (2017). Keynote 2: Cloud computing status and future. Panelist at 10th IEEE International Conference on Cloud Computing, Honolulu, HI.
• Elias, A.R., Golubovic, N., Krintz, C. & Wolski, R. (2017). Where’s the Bear? – Automating wildlife image processing using IoT and edge cloud systems. Proceedings of the Second International Conference on Internet-of-Things Design and Implementation – IoTDI.
• Jayathilaka, H., Krintz, C. & Wolski, R. (2017). Performance monitoring and root cause analysis for cloud-hosted web applications. Proceedings of the 26th International Conference on World Wide Web – WWW.
• Brevik, J. & Wolski, R. (2016). Providing statistical reliability guarantees in the AWS spot tier. Proceedings of the 24th High Performance Computing Symposium.