Edge computing has emerged as a new paradigm to bring cloud applications closer to users for increased performance. Unlike back-end cloud systems which consolidate their resources in a centralized data center location with virtually unlimited capacity, edge-clouds comprise distributed resources at various “computation spots”, each with very limited capacity. In this article, we consider Function-as-a-Service (FaaS) edge-clouds whereapplication providersdeploy their latency-critical functions to process user requests with strict response time deadlines. In this setting, we investigate the problem ofresource provisioningandallocation. After formulating the optimal solution, we propose resource allocation and provisioning algorithms across the spectrum of fully-centralized to fully-decentralized. We evaluate the performance of these algorithms in terms of their ability to utilize CPU resources and meet request deadlines under various system parameters. Our results indicate that practical decentralized strategies, which require no coordination among computation spots, achieve performance that is close to the optimal fully-centralized strategy with coordination overheads.

An increasing number of Low Latency Applications (LLAs) in the entertainment, IoT, and automotive domains require response times that challenge the traditional application provisioning using distant Data Centres. The fog computing paradigm extends cloud computing at the edge and middle-tier locations of the network, providing response times an order of magnitude smaller than those that can be achieved by the current “client-to-cloud” network model. Here, we address the challenges of provisioning heavily stateful LLA in the setting where fog infrastructure consists of third-party computing resources, i.e., cloudlets, that come in the form of “data centres in the box”. We introduce FogSpot, a charging mechanism for on-path, on-demand, application provisioning. In FogSpot, cloudlets offer their resources in the form of Virtual Machines (VMs) via markets, collocated with the cloudlets, that interact with forwarded users’ application requests for VMs in real time. FogSpot associates each cloudlet with a price based on applications’ demand. The proposed mechanism’s design takes into account the characteristics of cloudlets’ resources, such as their limited elasticity, and LLAs’ attributes, like their expected QoS gain and engagement duration. Lastly, FogSpot guarantees the end users’ requests truthfulness while focusing in maximising either each cloudlet’s revenue or resource utilisation.

A large number of emerging IoT applications rely on machine learning routines for analyzing data. Executing such tasks at the user devices improves response time and economizes network resources. However, due to power and computing limitations, the devices often cannot support such resource-intensive routines and fail to accurately execute the analytics. In this work, we propose to improve the performance of analytics by leveraging edge infrastructure. We devise an algorithm that enables the IoT devices to execute their routines locally; and then outsource them to cloudlet servers, only if they predict they will gain a significant performance improvement. It uses an approximate dual subgradient method, making minimal assumptions about the statistical properties of the system's parameters. Our analysis demonstrates that our proposed algorithm can intelligently leverage the cloudlet, adapting to the service requirements.

Decentralised cloud computing platforms enable individuals to offer and rent resources in a peer-to-peer fashion. They must assign resources from multiple sellers to multiple buyers and derive prices that match the interests and capacities of both parties. The assignment process must be decentralised, fair and transparent, but also protect the privacy of buyers.

Native applications running over handheld devices have an irreplaceable role in users' daily activities. That said, recent studies show that users download on average zero new applications on monthly basis, which suggests that new apps can face discoverability issues. In this work, we aim for a web-based, download/installation-free access to native application features through microservices (μ Services)that are shared between user devices in a peer-to-peer (P2P)manner. Such a P2P approach is self-scalable and requires no investment for μ Service deployment, unlike mobile edge computing or Data Centre. We introduce DEEM, a DEvice Edge Market design that enables device-hosted μServices to end-users. In DEEM, μ Service-based markets act as rendezvous points between available μ Service instances and clients. DEEM ensures the i) assignment of instances to the users that value them the most, in terms of QoS gain, and ii) devices' income maximisation. Our evaluation on synthetic settings demonstrates DEEM's capability in exploiting the pool of device instances for improving the application QoS in terms of latency.

Cloud computing has been tremendously successful in providing a commercial infrastructure for hosting computationally intensive applications. Nevertheless, an increasing number of Low Latency Applications (LLAs) notably in the entertainment, IoT, and automative domains require response times much smaller than the supported ones by the typical “client-to-cloud” network model. Cloudlets have been introduced as “data centres in a box”, for bringing computing resources “closer” to the end users. As a result, LLAs can take advantage of cloudlets to improve their Quality-of-Service (QoS) by reducing the underlying response times between their users and application instances' location. In this work, we study the emerging market of stateful LLAs' provisioning over geo-distributed third-party cloudlets. We assume that cloudlets offer their resources in the form of Virtual Machines (VMs) via collocated markets. Forwarding requests for LLAs interact with cloudlet markets for performing on-path and on-demand resource provisioning. We introduce a pricing scheme where users pay a fixed price for each time unit of their engagement to an LLA instance. Our evaluation on realistic topologies and application requests demonstrate the merits of on-demand provisioning when accompanied by a pay-as-you-go pricing scheme.

New and emerging applications in the entertainment (e.g., Virtual/Augmented Reality), IoT and automotive domains will soon demand response times an order of magnitude smaller than can be achieved by the current “client-to-cloud” network model. Edge-and Fog-computing have been proposed as the promise to deal with such extremely latency-sensitive applications. According to Edge-/Fog-Computing, computing resources are available at the edge of the network for applications to run their virtualised instances. We assume a distributed computing environment, where In-Network Computing Providers (IN CPs) deploy and lease edge resources, while Application Service Providers (AppSPs) have the opportunity to rent those resources to meet their application's latency demands. We build an auction-based resource allocation and provisioning mechanism which produces a map of application instances in the edge computing infrastructure (hence, acronymed Edge-MAP). Edge-MAP takes into account users' mobility (i.e., users connecting to different cell stations over time) and the limited computing resources available in edge micro-clouds to allocate resources to bidding applications. On the micro-level, Edge-MAP relies on Vickrey-English-Dutch (VED) auctions to perform robust resource allocation, while on the macro-level it fosters competition among neighbouring IN CPs. In contrast to related studies in the area, Edge-MAP can scale to any number of applications, adapt to dynamic network conditions rapidly and reallocate resources in polynomial time. Our evaluation demonstrates Edge-MAP's capability of taking into account the inherent challenges of the provisioning problem we consider.

Edge computing has emerged as a new paradigm to bring cloud applications closer to users for increased performance. ISPs have the opportunity to deploy private edge-clouds in their infrastructure to generate additional revenue by providing ultra-low latency applications to local users. We envision a rapid increase in the number of such applications for “edge” networks in the near future with virtual/augmented reality (VR/AR), networked gaming, wearable cognitive assistance, autonomous driving and IoT analytics having already been proposed for edge- clouds instead of the central clouds to improve performance. This raises new challenges as the complexity of the resource allocation problem for multiple services with latency deadlines (i.e., which service to place at which node of the edge-cloud in order to satisfy the latency constraints) becomes significant. In this paper, we propose a set of practical, uncoordinated strategies for service placement in edge-clouds. Through extensive simulations using both synthetic and real-world trace data, we demonstrate that uncoordinated strategies can perform comparatively well with the optimal placement solution, which satisfies the maximum amount of user requests.

In Delay Tolerant Networks (DTNs), there is a fundamental tradeoff between the aggregate transport cost of a packet and the delay in its delivery. We study this tradeoff in the context of geographical routing in wireless DTNs. We first specify the optimal cost/delay tradeoff, i.e., the tradeoff under optimal network operation, using a dynamic network construction termed the Cost/Delay Evolving Graph (C/DEG) and the Optimal Cost/Delay Curve (OC/DC), a function that gives the minimum possible aggregate transportation cost versus the maximum permitted delivery delay. We proceed to evaluate the performance of two known delay tolerant geographic routing rules, i.e., MOVE and AeroRP, a delay tolerant version of the geographic routing rule that selects as next relay the node for which the cost-per-progress ratio is minimized, and finally two novel rules, the Balanced Ratio Rule (BRR) and the Composite Rule (CR). The evaluation is in terms of the aggregate packet transmission cost as a function of the maximum permitted packet delivery delay. Simulations show that CR achieves a cost/delay tradeoff that is overall the closest to the optimal one specified by the OC/DC, while BRR achieves the smallest aggregate transmission costs for large packet delays and a fixed transmission cost model.