Umeå universitets logga

The emerging edge computing infrastructure provides distributed and heterogeneous resources closer to where data is generated and where end-users are located, thereby significantly reducing latency. With the recent advances in telecommunication systems, software architecture, and machine learning, there is a noticeable increase in applications that require processing times within tight latency constraints, i.e. latency-sensitive applications. For instance, numerous video analytics applications, such as traffic control systems, necessitate real-time processing capabilities. Orchestrating such applications at the edge offers numerous advantages, including lower latency, optimized bandwidth utilization, and enhanced scalability. However, despite its potential, effectively managing such latency-sensitive applications at the edge poses several challenges such as constrained compute resources, which holds back the full promise of edge computing.

This thesis proposes approaches to efficiently deploy latency-sensitive applications on the edge infrastructure. It partly addresses general applications with microservice architectures and party addresses the increasingly more important video analytics applications for the edge. To do so, this thesis proposes various application- and system-level solutions aiming to efficiently utilize constrained compute capacity on the edge while meeting prescribed latency constraints. These solutions primarily focus on effective resource management approaches and optimizing incoming workload inputs, considering the constrained compute capacity of edge resources. Additionally, the thesis explores the synergy effects of employing both application- and system-level resource optimization approaches together.

The results demonstrate  the effectiveness of the proposed solutions in enhancing the utilization of edge resources for latency-sensitive applications while adhering to application constraints. The proposed resource management solutions, alongside application-level optimization techniques, significantly improve resource efficiency while satisfying application requirements. Our results show that our solutions for microservice architectures significantly improve end-to-end latency by up to 800% while minimizing edge resource usage. Additionally, the results indicate that our application- and system-level optimizations for orchestrating edge resources for video analytics applications can increase the overall throughput by up to 60%. 

Numerous edge applications that rely on video analytics demand precise, low-latency processing of multiple video streams from cameras. When these cameras are mobile, such as when mounted on a car or a robot, the processing load on the shared edge GPU can vary considerably. Provisioning the edge with GPUs for the worst-case load can be expensive and, for many applications, not feasible. In this paper, we introduce RAVAS, a Real-time Adaptive stream Video Analytics System that enables efficient edge GPU sharing for processing streams from various mobile cameras. RAVAS uses Q-Learning to choose between a set of Deep Neural Network (DNN) models with varying accuracy and processing requirements based on the current GPU utilization and workload. RAVAS employs an innovative resource allocation strategy to mitigate interference during concurrent GPU execution. Compared to state-of-the-art approaches, our results show that RAVAS incurs 57% less compute overhead, achieves 41% improvement in latency, and 43% savings in total GPU usage for a single video stream. Processing multiple concurrent video streams results in up to 99% and 40% reductions in latency and overall GPU usage, respectively, while meeting the accuracy constraints.

Edge cloud systems reduce the latency between users and applications by offloading computations to a set of small-scale computing resources deployed at the edge of the network. However, since edge resources are constrained, they can become saturated and bottlenecked due to increased load, resulting in an exponential increase in response times or failures. In this paper, we argue that an application can be split between the edge and the cloud, allowing for better performance compared to full migration to the cloud, releasing precious resources at the edge. We model an application's internal call-Graph as a Directed-Acyclic-Graph. We use this model to develop MicroSplit, a tool for efficient splitting of microservices between constrained edge resources and large-scale distant backend clouds. MicroSplit analyzes the dependencies between the microservices of an application, and using the Louvain method for community detection---a popular algorithm from Network Science---decides how to split the microservices between the constrained edge and distant data centers. We test MicroSplit with four microservice based applications in various realistic cloud-edge settings. Our results show that Microsplit migrates up to 60% of the microservices of an application with a slight increase in the mean-response time compared to running on the edge, and a latency reduction of up to 800% compared to migrating the entire application to the cloud. Compared to other methods from the State-of-the-Art, MicroSplit reduces the total number of services on the edge by up to five times, with minimal reduction in response times.

Many edge applications rely on expensive Deep-Neural-Network (DNN) inference-based video analytics. Typically, a single instance of an inference service analyzes multiple real-time camera streams concurrently. In many cases,  only a fraction of these streams contain objects-of-interest at a given time. Hence, it is a waste of computational resources to process all frames from all cameras using the DNNs. On-camera filtration of frames has been suggested as a possible solution to improve the system efficiency and reduce resource wastage. However, many cameras do not have on-camera processing or filtering capabilities. In addition, filtration can be enhanced if frames across the different feeds are selected and prioritized for processing based on the system load and the available resource capacity. This paper introduces CVF, a Cross-video Filtration framework designed around video content and resource constraints. The CVF pipeline leverages compressed-domain data from encoded video formats, lightweight binary classification models, and an efficient prioritization algorithm. This enables the effective filtering of cross-camera frames from multiple sources, processing only a fraction of frames using resource-intensive DNN models. Our experiments show that CVF is capable of reducing the overall response time of video analytics pipelines by up to 50% compared to state-of-the-art solutions while increasing the throughput by up to 120%.