In this paper, we propose a novel agentic AI system called XPF, which enables users to create "agents" using just natural language, where each agent is capable of executing complex, real-world business workflows in an accurate and reliable manner. XPF provides an interface to develop and iterate over the agent creation process and then deploy the agent in production when satisfactory results are produced consistently. The key components of XPF include: (a) planner, which leverages LLM to generate a step-by-step plan, which can further be edited by a human (b) compiler, which leverages LLM to compile the plan into a flow graph (c) executor, which handles distributed execution of the flow graph (using LLM, tools, RAG, etc.) on an underlying cluster and (d) verifier, which helps in verification of the output (through human generated tests or auto-generated tests using LLM). We develop five different agents using XPF and conduct experiments to evaluate one particular aspect i.e. difference in accuracy and reliability of the five agents with "human-generated" vs "auto-generated" plans. Our experiments show that we can get much more accurate and reliable response for a business workflow when step-by-step instructions (in natural language) are given by a human familiar with the workflow, rather than letting the LLM figure out the execution plan steps. In particular, we observe that "human-generated" plan almost always gives 100% accuracy whereas "auto-generated" plan almost never gives 100% accuracy. In terms of reliability, we observe through Rouge-L, Blue and Meteor scores, that the output from "human-generated" plan is much more reliable than "auto-generated" plan.

In Video Analytics Pipelines (VAP), Analytics Units (AUs) such as object detection and face recognition operating on remote servers rely heavily on surveillance cameras to capture high-quality video streams to achieve high accuracy. Modern network cameras offer an array of parameters that directly influence video quality. While a few of such parameters, e.g., exposure, focus and white balance, are automatically adjusted by the camera internally, the others are not. We denote such camera parameters as non-automated (NAUTO) parameters. In this work, we first show that in a typical surveillance camera deployment, environmental condition changes can have significant adverse effect on the accuracy of insights from the AUs, but such adverse impact can potentially be mitigated by dynamically adjusting NAUTO camera parameters in response to changes in environmental conditions. Second, since most end-users lack the skill or understanding to appropriately configure these parameters and typically use a fixed parameter setting, we presentCamTuner, to our knowledge, the first framework that dynamically adapts NAUTO camera parameters to optimize the accuracy of AUs in a VAP in response to adverse changes in environmental conditions.CamTuneris based on SARSA reinforcement learning and it incorporates two novel components: a light-weight analytics quality estimator and a virtual camera that drastically speed up offline RL training. Our controlled experiments and real-world VAP deployment show that compared to a VAP using the default camera setting,CamTunerenhances VAP accuracy by detecting 15.9% additional persons and 2.6% –4.2% additional cars (without any false positives) in a large enterprise parking lot.CamTuneropens up new avenues for elevating video analytics accuracy, transcending mere incremental enhancements achieved through refining deep-learning models.

Edge computing has emerged as a transformative technology that reduces application latency, improves cost efficiency, enhances security, and enables large-scale deployment of applications across various domains. In environmental monitoring, systems such as MegaSense[49], use low-cost sensors to gather and process real-time air quality data through edge-cloud collaboration, highlighting the critical role of edge computing in enabling scalable, efficient solutions. Similarly, marine science increasingly requires real-time processing and analysis of marine data from remote, resource-constrained environments. In this paper, we extend the power of edge computing by integrating it with Generative Artificial Intelligence(GenAI),specifically large language models (LLMs), to address challenges in marine science applications. We propose DiCE-M (Distributed Code generation and Execution for Marine applications), a robust system that uses LLM to generate distributed code for marine applications and then utilizes a runtime to efficiently execute it on an edge+cloud computing infrastructure. Specifically, DiCE-M leverages edge computing to execute lightweight AI models locally on unmanned surface vehicles(USVs)while offloading complex tasks to the cloud, thus balancing computational load and enabling realtime monitoring in marine environments. We use marine litter identification as an example application to demonstrate the utility of DiCE-M. Our results show that DiCE-M reduces latency by more than 2X when marine litter is not detected and cuts cloud computing costs by more than half compared to traditional cloud-based approaches. By selectively cropping and transmitting relevant image portions, DiCE-M further improves bandwidth efficiency, making it a reliable and cost-effective solution for deploying AI-drivenapplications on resource-constrained USVs in dynamic marine environments.

One of the key metrics of interest for stream processing applications is “latency”, which indicates the total time it takes for the application to process and generate insights from streaming input data. For mission-critical video analytics applications like surveillance and monitoring, it is of paramount importance to report an incident as soon as it occurs so that necessary actions can be taken right away. Stream processing applications are typically developed as a chain of microser-vices and are deployed on container orchestration platforms like Kubernetes. Allocation of system resources like “cpu” and “memory” to individual application microservices has direct impact on “latency”. Kubernetes does provide ways to allocate these resources e.g. through fixed resource allocation or through vertical pod autoscaler (VPA), however there is no straight-forward way in Kubernetes to prioritize “latency” for an end-to-end application pipeline. In this paper, we present LARA, which is specifically designed to improve “latency” of stream processing application pipelines. LARA uses a regression-based technique for resource allocation to individual microservices. We implement four real-world video analytics application pipelines i.e. license plate recognition, face recognition, human attributes detection and pose detection, and show that compared to fixed allocation, LARA is able to reduce latency by up to 2.8X and is consistently better than VPA. While reducing latency, LARA is also able to deliver over 2X throughput compared to fixed allocation and is almost always better than VPA.

Modern applications are designed as an interacting set of microservices, and these applications are typically deployed on container orchestration platforms like Kubernetes. Several attractive features in Kubernetes make it a popular choice for deploying applications, and automatic scaling is one such feature. The default horizontal scaling technique in Kubernetes is the Horizontal Pod Autoscaler (HPA). It scales each microservice independently while ignoring the interactions among the microservices in an application. In this paper, we show that ignoring such interactions by HPA leads to inefficient scaling, and the optimal scaling of different microservices in the application varies as the stream content changes. To automatically adapt to variations in stream content, we present a novel system called DataX AutoScaler that leverages knowledge of the entire stream processing application pipeline to efficiently auto-scale different microservices by taking into account their complex interactions. Through experiments on real-world video analytics applications, such as face recognition and pose classification, we show that DataX AutoScaler adapts to variations in stream content and achieves up to 43% improvement in overall application performance compared to a baseline system that uses HPA.

IoT sensors, especially video cameras, are ubiquitously deployed around the world to perform a variety of computer vision tasks in several verticals including retail, health-care, safety and security, transportation, manufacturing, etc. To amortize their high deployment effort and cost, it is desirable to perform multiple video analytics tasks, which we refer to as Analytical Units (AUs), off the video feed coming out of every camera. As AUs typically use deep-learning based AI/ML models, their performances depend on the quality of the input video. The most recent work has shown that dynamically adjusting the camera setting exposed by popular network cameras can help improve the quality of the video feed and hence the AU accuracy, in a single AU setting. In this paper, we first show that in a multi-AU setting, changing the camera setting has disproportionate impact on different AUs performance. In particular, the optimal setting for one AU may severely degrade the performance for another AU, and further, the impact on different AUs varies as the environmental condition changes. We then present Elixir, a system to enhance the video stream quality for multiple analytics on a video stream. Elixir leverages Multi-Objective Reinforcement Learning (MORL), where the RL agent caters to the objectives from different AUs and adjusts the camera setting to simultaneously enhance the performance of all AUs. To define the multiple objectives in MORL, we develop new AU-specific quality estimator values for each individual AU. We evaluate Elixir through real-world experiments on a testbed with three cameras deployed next to each other (overlooking a large enterprise parking lot) running Elixir and two baseline approaches, respectively. Elixir correctly detects 7.1% (22,068) and 5.0% (15,731) more cars, 94% (551) and 72% (478) more faces, and 670.4% (4975) and 158.6% (3507) more persons than the default-setting and time-sharing approaches, respectively. It also detects 115 license plates, far more than the time-sharing approach (7) and the default setting (0).

We present "Application in a Box" (AnB) product concept aimed at simplifying the deployment and operation of remote 5G applications. AnB comes pre-configured with all necessary hardware and software components, including sensors like cameras, hardware and software components for a local 5G wireless network, and 5G-ready apps. Enterprises can easily download additional apps from an App Store. Setting up a 5G infrastructure and running applications on it is a significant challenge, but AnB is designed to make it fast, convenient, and easy, even for those without extensive knowledge of software, computers, wireless networks, or AI-based analytics. With AnB, customers only need to open the box, set up the sensors, turn on the 5G networking and edge computing devices, and start running their applications. Our system software automatically deploys and optimizes the pipeline of microservices in the application on a tiered computing infrastructure that includes device, edge, and cloud computing. Application scalability, dynamic resource management, placement of critical tasks for low-latency response, and dynamic network bandwidth allocation for efficient 5G network usage are all automatically orchestrated.AnB offers cost savings, simplified setup and management, and increased reliability and security. We’ve implemented several real-world applications, such as collision prediction at busy traffic light intersections and remote construction site monitoring using video analytics. With AnB, deployment and optimization effort can be reduced from several months to just a few minutes. This is the first-of-its-kind approach to easing deployment effort and automating self-optimization of the application during system operation.

With the growth of 5G, Internet of Things (IoT), edge computing and cloud computing technologies, the infrastructure (compute and network) available to emerging applications (AR/VR, autonomous driving, industry 4.0, etc.) has become quite complex. There are multiple tiers of computing (IoT devices, near edge, far edge, cloud, etc.) that are connected with different types of networking technologies (LAN, LTE, 5G, MAN, WAN, etc.). Deployment and management of applications in such an environment is quite challenging. In this paper, we propose ROMA, which performs resource orchestration for microservices-based 5G applications in a dynamic, heterogeneous, multi-tiered compute and network fabric. We assume that only application-level requirements are known, and the detailed requirements of the individual microservices in the application are not specified. As part of our solution, ROMA identifies and leverages the coupling relationship between compute and network usage for various microservices and solves an optimization problem in order to appropriately identify how each microservice should be deployed in the complex, multi-tiered compute and network fabric, so that the end-to-end application requirements are optimally met. We implemented two real-world 5G applications in video surveillance and intelligent transportation system (ITS) domains. Through extensive experiments, we show that ROMA is able to save up to 90%, 55% and 44% compute and up to 80%, 95% and 75% network bandwidth for the surveillance (watchlist) and transportation application (person and car detection), respectively. This improvement is achieved while honoring the application performance requirements, and it is over an alternative scheme that employs a static and overprovisioned resource allocation strategy by ignoring the resource coupling relationships.

In Video Analytics Pipelines (VAP), Analytics Units (AUs) such as object detection and face recognition running on remote servers critically rely on surveillance cameras to capture high-quality video streams in order to achieve high accuracy. Modern IP cameras come with a large number of camera parameters that directly affect the quality of the video stream capture. While a few of such parameters, e.g., exposure, focus, white balance are automatically adjusted by the camera internally, the remaining ones are not. We denote such camera parameters as non-automated (NAUTO) parameters. In this paper, we first show that environmental condition changes can have significant adverse effect on the accuracy of insights from the AUs, but such adverse impact can potentially be mitigated by dynamically adjusting NAUTO camera parameters in response to changes in environmental conditions. We then present CamTuner, to our knowledge, the first framework that dynamically adapts NAUTO camera parameters to optimize the accuracy of AUs in a VAP in response to adverse changes in environmental conditions. CamTuner is based on SARSA reinforcement learning and it incorporates two novel components: a light-weight analytics quality estimator and a virtual camera that drastically speed up offline RL training. Our controlled experiments and real-world VAP deployment show that compared to a VAP using the default camera setting, CamTuner enhances VAP accuracy by detecting 15.9% additional persons and 2.6%--4.2% additional cars (without any false positives) in a large enterprise parking lot and 9.7% additional cars in a 5G smart traffic intersection scenario, which enables a new usecase of accurate and reliable automatic vehicle collision prediction (AVCP). CamTuner opens doors for new ways to significantly enhance video analytics accuracy beyond incremental improvements from refining deep-learning models.

Centralized cloud computing with 100+ milliseconds network latencies cannot meet the tens of milliseconds to sub-millisecond response times required for emerging 5G applications like autonomous driving, smart manufacturing, tactile internet, and augmented or virtual reality. We describe a new, dynamic runtime that enables such applications to make effective use of a 5G network, computing at the edge of this network, and resources in the centralized cloud, at all times. Our runtime continuously monitors the interaction among the microservices, estimates the data produced and exchanged among the microservices, and uses a novel graph min-cut algorithm to dynamically map the microservices to the edge or the cloud to satisfy application-specific response times. Our runtime also handles temporary network partitions, and maintains data consistency across the distributed fabric by using microservice proxies to reduce WAN bandwidth by an order of magnitude, all in an application-specific manner by leveraging knowledge about the application's functions, latency-critical pipelines and intermediate data. We illustrate the use of our runtime by successfully mapping two complex, representative real-world video analytics applications to the AWS/Verizon Wavelength edge-cloud architecture, and improving application response times by 2x when compared with a static edge-cloud implementation.

Microservices-based video analytics pipelines routinely use multiple deep convolutional neural networks. We observe that the best allocation of resources to deep learning engines (or microservices) in a pipeline, and the best configuration of parameters for each engine vary over time, often at a timescale of minutes or even seconds based on the dynamic content in the video. We leverage these observations to develop Magic-Pipe, a self-optimizing video analytic pipeline that leverages AI techniques to periodically self-optimize. First, we propose a new, adaptive resource allocation technique to dynamically balance the resource usage of different microservices, based on dynamic video content. Then, we propose an adaptive microservice parameter tuning technique to balance the accuracy and performance of a microservice, also based on video content. Finally, we propose two different approaches to reduce unnecessary computations due to unavoidable mismatch of independently designed, re-usable deep-learning engines: a deep learning approach to improve the feature extractor performance by filtering inputs for which no features can be extracted, and a low-overhead graph-theoretic approach to minimize redundant computations across frames. Our evaluation of Magic-Pipe shows that pipelines augmented with self-optimizing capability exhibit application response times that are an order of magnitude better than the original pipelines, while using the same hardware resources, and achieving similar high accuracy.

Identification of people with elevated body temperature can reduce or dramatically slow down the spread of infectious diseases like COVID-19. We present a novel fever-screening system, F3S, that uses edge machine learning techniques to accurately measure core body temperatures of multiple individuals in a free-flow setting. F3S performs real-time sensor fusion of visual camera with thermal camera data streams to detect elevated body temperature, and it has several unique features: (a) visual and thermal streams represent very different modalities, and we dynamically associate semantically-equivalent regions across visual and thermal frames by using a new, dynamic alignment technique that analyzes content and context in real-time, (b) we track people through occlusions, identify the eye (inner canthus), forehead, face and head regions where possible, and provide an accurate temperature reading by using a prioritized refinement algorithm, and (c) we robustly detect elevated body temperature even in the presence of personal protective equipment like masks, or sunglasses or hats, all of which can be affected by hot weather and lead to spurious temperature readings. F3S has been deployed at over a dozen large commercial establishments, providing contact-less, free-flow, real-time fever screening for thousands of employees and customers in indoors and outdoor settings.

GPU-based clusters are increasingly being deployed in HPC environments to accelerate a variety of scientific applications. Despite their growing popularity, the GPU devices themselves are under-utilized even for many computationally-intensive jobs. This stems from the fact that the typical GPU usage model is one in which a host processor periodically offloads computationally intensive portions of an application to the coprocessor. Since some portions of code cannot be offloaded to the GPU (for example, code performing network communication in MPI applications), this usage model results in periods of time when the GPU is idle. GPUs could be time-shared across jobs to "fill" these idle periods, but unlike CPU resources such as the cache, the effects of sharing the GPU are not well understood. Specifically, two jobs that time-share a single GPU will experience resource contention and interfere with each other. The resulting slow-down could lead to missed job deadlines. Current cluster managers do not support GPU-sharing, but instead dedicate GPUs to a job for the job's lifetime.

Coprocessors such as GPUs are increasingly being deployed in clusters to process scientific and compute-intensive jobs. In this work, we study if GPU-based heterogeneous clusters can benefit client-server applications. Specifically, we consider the practical situation where multiple client-server applications share a heterogeneous cluster (multi-tenancy), and experience unpredictable variations in incoming client request rates, including steep load spikes. Even for "compute-intensive" client-server applications, it is unclear if a GPU-based cluster can seamlessly deliver acceptable response times in the presence of multi-tenancy and load spikes. We argue that a cluster-level scheduler that is aware of application load, request deadlines and the heterogeneity is necessary in this situation. We propose a novel scheduler called Symphony that enables efficient, dynamic sharing of a GPU-based heterogeneous cluster across multiple concurrently-executing client-server applications, each with arbitrary load spikes. Symphony performs three key tasks: it (i) monitors the load on each application, (ii) collects past performance data and dynamically builds simple performance models of available processing resources and (iii) computes a priority for pending requests based on the above parameters and the requests' slack. Based on this, it reorders client requests across different applications to achieve acceptable response times. We also define how client-server applications should interact with a scheduler such as Symphony, and develop an API to this end. We deploy Symphony as user-space middleware on a high-end heterogeneous cluster with dual quad-core Xeon CPUs and dual NVIDIA Fermi GPUs. An evaluation using representative applications shows that in the presence of load spikes (i) Symphony incurs 2-20× fewer requests that do not meet response time constraints compared with other schedulers, and (ii) in order to achieve the same performance as Symphony, other schedulers need 2× more cluster nodes.