Mobile Stream Processing (MSP) provides a promising approach to run computation-intensive stream applications, e.g., video face recognition, on a cluster of mobile devices at the edge. However, the performance of MSP is severely restricted by the fluctuating bandwidth and intermittent connectivity of the wireless networks connecting those devices. Therefore, to achieve a good MSP performance, implementing a resilient MSP system that adapts to dynamic edge networks is essential. In this paper, we present R-MStorm, a resilient MSP system deployed at the edge. R-MStorm improves the system survivability by (1) assigning tasks to mobile devices with higher availability to improve the availability of whole system; (2) assigning tasks of the same application components to different devices to increase the diversity of physical stream paths. Besides, to efficiently divide the output of upstream tasks to downstream tasks, R-MStorm adopts adaptive stream grouping, which considers both the transmission rate to and processing rate at each downstream task. Moreover, to alleviate congestion caused by network disconnection and stream redirection, adaptive stream selection is applied to skip some data to achieve a short response time. We conduct extensive experiments on R-MStorm by executing a video face recognition App under different network conditions. The experimental results show that, compared with baseline approaches, R-MStorm achieves up to 1.5x higher throughput, 75% lower response time, at a cost of 3.3% accuracy loss.
By leveraging resources from the Fed4Fire+ City-Lab testbed, we design the PiGeon edge computing platform that experiments solution that enable ICN based edge services in wireless mesh networks (WMNs). PiGeon combines into a platform several trends in edge computing namely the ICN (Information-Centric Networking), the containerization of services exemplified by Docker, novel service placement algorithms and the increasing availability of energy efficient but still powerful hardware at user premises (Raspberry Pi, mini-PCs, and enhanced home gateways). We underpin the PiGeon platform with Docker container-based service that can be seamlessly delivered, cached and deployed at the network edge. The core of the PiGeon platform is the Decision Engine making a decision on where and when to deploy a service instance to satisfy the service requirements while considering the network status and available hardware resources.We collect network data from a real citywide mesh network such as CityLab FIRE testbed located at the city of Antwerp, Belgium. The collected data is used to feed our service placement heuristic within the PiGeon platform. Through a real deployment in CityLab testbed, we show that our service placement heuristic improves the response time up to 37% for stateful services (Web2.0 service). Apart from improving the QoS for end-users, our results show that ICN plays a key role in improving the service delivery time as well as reducing the traffic consumption in WMNs. The overall effect of ICN in our platform is that most content and service delivery requests can be satisfied very close to the client device, many times just one hop away, decoupling QoS from intra-network traffic and origin server load.
Residents in cities typically use third-party platforms such as Google Maps for route planning services. While providing near real-time processing, these state of the art centralized deployments are limited to multiprocessing environments in data centers. This raises privacy concerns, increases risk for critical data and causes vulnerability to network failure. In this paper, we propose to use decentralized road side units (RSU) (owned by the city) to perform route planning. We divide the city road network into grids, each assigned an RSU where traffic data is kept locally, increasing security and resiliency such that the system can perform even if some RSUs fail. Route generation is done in two steps. First, an optimal grid sequence is generated, prioritizing shortest path calculation accuracy but not RSU load. Second, we assign route planning tasks to the grids in the sequence. Keeping in mind RSU load and constraints, tasks can be allocated and executed in any non-optimal grid but with lower accuracy. We evaluate this system using Metropolitan Nashville road traffic data. We divided the area into 613 grids, configuring load and neighborhood sizes to meet delay constraints while maximizing model accuracy. The results show that there is a 30% decrease in processing time with a decrease in model accuracy of 99% to 92.3%, by simply increasing the search area to the optimal grid's immediate neighborhood.
Today, communication between IoT devices heavily relies on fog-based publish/subscribe (pub/sub) systems. Communicating via the cloud, however, results in a latency that is too high for many IoT applications. In this paper, we describe the design of a fog-based pub/sub system that integrates edge resources to improve communication latency between end devices in proximity. To this end, geo-distributed broker instances organize themselves in dynamically sized broadcast groups. Each broadcast group comprises a set of well connected edge brokers that communicate directly using flooding. This minimizes communication latency and copes well with frequently updated subscriptions and mobile end devices, which is required by many IoT applications. Messages between broadcast groups are routed via a massively scalable fog broker that pre-filters messages to reduce excess data dissemination. Our approach, therefore, manages the tradeoff between latency and excess data.
Augmented reality applications are computationally intensive and have latency requirements in the range of 15-20 milliseconds. Fog computing addresses these requirements by providing on-demand computing capacity and lower latency by bringing the computational resources closer to the augmented reality devices. In this paper, we reviewed papers providing custom solutions for augmented reality using the fog architecture and identified that the ongoing research trends towards balancing quality-of-experience, energy, and latency for both single and collaborative multi-device augmented reality applications. Furthermore, some works also focus on providing architectures for fog-based augmented reality systems and also on the training of machine learning algorithms in the fog layers to improve user experience. Based on these findings, we provide some challenges and research directions that can facilitate the adoption of fog-based augmented reality systems.
The Function-as-a-Service (FaaS) model is a great fit for data and event processing in the Internet of Things (IoT). Sending all data to a cloud-based FaaS platform, however, may cause performance and privacy issues. While these issues could be mitigated using edge computing, existing FaaS approaches, designed for the cloud, are too heavyweight to run on small, constrained edge nodes.In this paper, we propose tinyFaaS, a new FaaS system that is specifically designed for edge environments and their unique challenges. Our platform is lightweight enough to run on low-performance single machine edge nodes, provides a CoAP endpoint to support communication with low-power devices, and uses Docker containers to isolate tenants. We evaluate tinyFaaS through a proof-of-concept implementation that we benchmark and compare to state-of-the-art FaaS platforms. For IoT processing scenarios, we find that tinyFaaS outperforms existing systems by at least an order of magnitude.
In recent years, MTConnect has emerged as a potential communication standard for cyber manufacturing (CM) domain by establishing remote monitoring of manufacturing processes through XML based data reporting structure and RESTful services. This paper presents the development and optimization of an MTConnect based edge computing node in service-oriented CM systems by utilizing data caching and processing at the edge. Unlike MTConnect agents, proposed MTConnect Edge Nodes (MENs) not only convert collected machining data to XML messages and host RESTful services, but also perform as an edge node by adopting a “hold-until-changed” approach for deciding which data to store and by keeping tracks of previously transmitted data to its clients to determine which data to transmit to whom and when. The primary objective is to minimize data storage requirements and cost while enabling rapid transmission and low bandwidth usage without increasing information loss. This paper describes the architecture of an MEN and its data caching and transmission strategies in details. Experiments were conducted in a CM testbed with three machine tools, raspberry pis hosting MENs and MTConnect agents, and two client applications to evaluate MEN’s performance with respect to the conventional approach. Results showed 96.2 percent reduction in required storage size with 52.5 percent reduction in average communication latency and 99.5 percent reduction in average message size for proposed MEN in different manufacturing scenarios.
In this study, the authors adopt the non-orthogonal multiple access (NOMA) technique to improve the spectrum efficiency in the wireless backhaul networks, whereby the downlink and uplink NOMA techniques are applied for the backhaul and access links, respectively. Due to the coupling between the backhaul and access transmission stages, the transmission power should be carefully allocated in both stages so that the overall throughput can be maximised. They start with the single user equipment (UE) case and consider different scenarios and analyse the tradeoff between the access and backhaul links, and the optimal power allocation solutions are obtained accordingly. They then extend the analysis to the multi-UE case and formulate the optimal power allocation problem, which is solved using the Lagrangian dual decomposition algorithm. Simulation results demonstrate that the proposed schemes are effective in improving the throughput and outperforms the conventional orthogonal multiple access technique under different network settings.
We present FLIC, a distributed software data caching framework for fogs that reduces network traffic and latency. FLICis targeted toward city-scale deployments of cooperative IoT devices in which each node gathers and shares data with surrounding devices. As machine learning and other data processing techniques that require large volumes of training data are ported to low-cost and low-power IoT systems, we expect that data analysis will be moved away from the cloud. Separation from the cloud will reduce reliance on power-hungry centralized cloud-based infrastructure. However, city-scale deployments of cooperative IoT devices often connect to the Internet with cellular service, in which service charges are proportional to network usage. IoT system architects must be clever in order to keep costs down in these scenarios. To reduce the network bandwidth required to operate city-scale deployments of cooperative IoT systems, FLIC implements a distributed cache on the IoT nodes in the fog. FLIC allows the IoT network to share its data without repetitively interacting with a simple cloud storage service reducing calls out to a backing store. Our results displayed a less than 2% miss rate on reads. Thus, allowing for only 5% of requests needing the backing store. We were also able to achieve more than 50% reduction in bytes transmitted per second.
The Function-as-a-Service (FaaS) paradigm has a lot of potential as a computing model for fog environments comprising both cloud and edge nodes. When the request rate exceeds capacity limits at the edge, some functions need to be offloaded from the edge towards the cloud.In this position paper, we propose an auction-based approach in which application developers bid on resources. This allows fog nodes to make a local decision about which functions to offload while maximizing revenue. For a first evaluation of our approach, we use simulation.
The emergence of real-time and data-intensive applications empowered by mobile computing and IoT devices is challenging the success of centralized data centers, and fostering the adoption of the paradigm of fog/edge computing. Differently from cloud data centers, fog nodes are geographically distributed in proximity to data prosumers, taking advantage of the emerging wireless communication technologies and mobile networks. The limited resources of densely distributed fog nodes call for their efficient use by hosted applications and services. To address this challenge, and the needs of different application scenarios, this paper proposes a serverless platform for edge computing. It starts motivating the adoption of a serverless architecture. Then, it presents the services and mechanisms that are the building blocks of a Serverless Edge Platform. The paper also proposes a prototype platform and its assessment. Obtained results demonstrate the feasibility of the proposed solution for satisfying different application requirements in diverse deployment configurations of heterogeneous fog nodes.
Detailed knowledge about the electrical power consumption in industrial production environments is a prerequisite to reduce and optimize their power consumption.Today's industrial production sites are equipped with a variety of sensors that, inter alia, monitor electrical power consumption in detail.However, these environments often lack an automated data collation and analysis.We present a system architecture that integrates different sensors and analyzes and visualizes the power consumption of devices, machines, and production plants.It is designed with a focus on scalability to support production environments of various sizes and to handle varying loads.We argue that a scalable architecture in this context must meet requirements for fault tolerance, extensibility, real-time data processing, and resource efficiency.As a solution, we propose a microservicebased architecture augmented by big data and stream processing techniques.Applying the fog computing paradigm, parts of it are deployed in an elastic, central cloud while other parts run directly, decentralized in the production environment.A prototype implementation of this architecture presents solutions how different kinds of sensors can be integrated and their measurements can be continuously aggregated.In order to make analyzed data comprehensible, it features a single-page web application that provides different forms of data visualization.We deploy this pilot implementation in the data center of a medium-sized enterprise, where we successfully monitor the power consumption of 16 servers.Furthermore, we show the scalability of our architecture with 20,000 simulated sensors.
We present models of Energy Packet Net-works with a single and multiple class of customers. Such models were recently developed to study the interactions between IT capabilities and its energy consumption. These models are based on a discrete model of energy (the so called Energy Packets), some assumptions on the stochastic durations of energy production and consumption and the explicit description of the interactions between energy needed for computations or data transmission. These stochastic models have, under some classical assumptions on the arrivals processes, a steady-state distribution which has a product form. Such a closed form solution allows to optimize the design of the system and balance between losses, delay and energy.
Power-of-random choices is a well-known phenomenon exploited in load balancing algorithms to achieve an extraordinary improvement at low cost. These algorithms are particularly suitable for the Fog computing model, since they don't require coordination among different fog nodes when they decide to mutually share their resources. In this paper, we propose LL(F, T ) a power-of-random choices based distributed peer-to-peer load balancing algorithm running on a set of autonomous cooperating fog nodes, where F is the protocol fan-out and T a threshold. Nodes implement a random choice over F fog nodes when their current load is above T. Through a mathematical analysis and preliminary simulations we show that tuning T very close to the node saturation condition, this algorithm achieves practically the same performance of its classical implementation requiring a single global scheduler, without the need for each and every job execution to be preceded by a time costly probing phase, a clear benefit for the low delay requirement of fog applications.
Fog computing is an emerging computing paradigm that uses processing and storage capabilities located at the edge, in the cloud, and possibly in between. Testing fog applications, however, is hard since runtime infrastructures will typically be in use or may not exist, yet. In this paper, we propose an approach that emulates such infrastructures in the cloud. Developers can freely design emulated fog infrastructures, configure their performance characteristics, and inject failures at runtime to evaluate their application in various deployments and failure scenarios. We also present our proof-of-concept implementation MockFog and show that application performance is comparable when running on MockFog or a small fog infrastructure testbed.
Internet of Things is nowadays growing faster than ever before. Operators are planning or already creating dedicated networks for this type of devices. There is a need to create dedicated solutions for this type of network, especially solutions related to information security. In this article we present a mechanism of security-aware routing, which takes into account the evaluation of trust in devices and packet flows. We use trust relationships between flows and network nodes to create secure SDN paths, not ignoring also QoS and energy criteria. The system uses SDN infrastructure, enriched with Cognitive Packet Networks (CPN) mechanisms. Routing decisions are made by Random Neural Networks, trained with data fetched with Cognitive Packets. The proposed network architecture, implementing the security-by-design concept, was designed and is being implemented within the SerIoT project to demonstrate secure networks for the Internet of Things (IoT).
Multi-access Edge Computing (MEC) is emerging as the next evolution in computing infrastructure. However, the broad adoption of edge computing solutions is still lagging behind. We posit that a significant factor contributing to this trend is related to the challenges in evaluating the trade-offs offered by edge computing and answering a simple question: Will the edge investments be offset with commensurate gains in performance metrics such as improved application responsiveness or lower backhaul bandwidth costs? Existing simulation tools, developed for cloud computing and networking research, require a large amount of effort to be applied to the edge computing space. Their native abstractions and interfaces pose limitations in how such frameworks can be extended to allow for characterization and analysis of diverse edge computing use cases. Moreover, as we demonstrate, they can be slow, taking days, to provide insights. To address this gap, we develop Castnet - a fast, extensible, and easy-to-use framework for evaluating the benefits of MEC for different edge deployment models, configuration and functions. We demonstrate that Castnet leads to significant gains in the time-to-insight while achieving the same fidelity in the results as existing tools. We illustrate the utility of Castnet using two scenarios: an investigation of the impact of edge storage capacity for edge-based content caches, and the impact of edge location for an edge-based intrusion detection service.
Numerous recent research efforts have been devoted to edge computing due to its key role in supporting emerging IoT applications. Prior to deploying edge technologies to real-world environments, they need to be adequately tested, tuned and validated. However, to our best knowledge, a testing platform for edge computing that provides both networking and computing realism at low cost is still missing. In this paper, we present EmuEdge, a hybrid emulator based on Linux netns and Xen for full-stack edge computing emulation. Supporting both container and VM, EmuEdge is the first edge computing emulator that takes advantage of both OS level and system level virtualization. Meanwhile, EmuEdge provides interfaces to integrate with both simulators and testbeds. The hybrid design provides on-demand realism on both networking and computing plane while ensuring emulation scalability. Moreover, it supports network replay and can be configured with an easy-to-use configuration tool. Through extensive experiments, we demonstrate that EmuEdge provides more realistic computation isolation and higher network fidelity than the state-of-the-art solutions.
Data Protection is a major research topic concerning the Internet of Things (IoT). IoT systems continue to permeate deeper into our personal lives, where devices sense, process, and store all kinds of data. This poses various challenges to security and privacy aspects, especially to applications running on resource constrained devices. In this paper we evaluate selected, well established data protection mechanisms that enable confidentiality and integrity of data. Specifically, we look into the performance of different cryptographic block and stream ciphers, hashing algorithms, message authentication codes, signature mechanisms, and key exchange protocols executed on state-of-the-art resource constrained devices. By providing limitations and data throughput values, our obtained results ease the calculation of performance/data protection thresholds and facilitate the design and development of secure IoT systems.
Fog computing brings the convenience of cloud computing closer to the edge of the Internet and even further into local area networks. A central aspect of fog computing in an industrial scenario concerns the orchestration of applications to be executed on an existing automation system network. This paper studies the problem of mapping distributed applications onto an industrial fog network so as to minimize data transfer cost while adhering to resource constraints. The resulting fog application allocation problem is NP-complete and can be described as an integer linear program. We present and evaluate three polynomial-time heuristics, which exploit the problem structure of a typical fog application and are shown to find high quality allocations efficiently. Through extensive simulations on practically relevant scenarios, we show that the best performing heuristic has performance comparable to the optimal solution across most tested problem instances. The experiments with a real industrial fog application show that the best performing heuristic has an optimality gap of only 3.6%.
