Emerging technologies and applications including Internet of Things, social networking, crowd-sourcing generate large amounts data at the network edge. Machine learning models are often built from collected data, to enable detection, classification, prediction future events. Due bandwidth, storage, privacy concerns, it is impractical send all a centralized location. In this paper, we consider problem model parameters distributed across multiple edge nodes, without sending raw place. Our...

Emerging technologies and applications including Internet of Things (IoT), social networking, crowd-sourcing generate large amounts data at the network edge. Machine learning models are often built from collected data, to enable detection, classification, prediction future events. Due bandwidth, storage, privacy concerns, it is impractical send all a centralized location. In this paper, we consider problem model parameters distributed across multiple edge nodes, without sending raw place....

We study the dynamic service migration problem in mobile edge-clouds that host cloud-based services at network edge. This offers benefits of reduction overhead and latency but requires migrations as user locations change over time. It is challenging to make these decisions an optimal manner because uncertainty node mobility well possible non-linearity transmission costs. In this paper, we formulate a sequential decision making for using framework Markov Decision Process (MDP). Our...

Mobile micro-clouds are promising for enabling performance-critical cloud applications. However, one challenge therein is the dynamics at network edge. In this paper, we study how to place service instances cope with these dynamics, where multiple users and coexist in system. Our goal find optimal placement (configuration) of minimize average cost over time, leveraging ability predicting future parameters known accuracy. We first propose an offline algorithm that solves configuration a...

In mobile edge computing, local servers can host cloud-based services, which reduces network overhead and latency but requires service migrations as users move to new locations. It is challenging make migration decisions optimally because of the uncertainty in such a dynamic cloud environment. this paper, we formulate problem Markov decision process (MDP). Our formulation captures general cost models provides mathematical framework design optimal policies. order overcome complexity...

Mobile edge computing allows wireless users to exploit the power of cloud without large communication delay. To serve data-intensive applications (e.g., augmented reality, video analytics) from edge, we need, in addition CPU cycles and memory for computation, storage resource storing server data network bandwidth receiving user-provided data. Moreover, placement needs be adapted over time time-varying demands, while considering system stability operation cost. We address this problem by...

Mobile edge computing provides the opportunity for wireless users to exploit power of cloud without a large communication delay. To serve data-intensive applications (e.g., video analytics, machine learning tasks) from edge, we need, in addition computation resources, storage resources storing server code and data as well network bandwidth receiving user-provided data. Moreover, due time-varying demands, placement needs be adjusted over time, which raises concerns system stability operation...

Many systems or applications have been developed for distributed environments with the goal of attaining multiple objectives in face environmental challenges such as high dynamics/hostility severe resource constraints (e.g., energy communications bandwidth). Often are conflicting each other, requiring optimal tradeoff analyses between objectives. This paper is mainly concerned how to model a system and optimize their performance. We first conduct comprehensive survey state-of-the-art...

Mobile micro-cloud is an emerging technology in distributed computing, which aimed at providing seamless computing/data access to the edge of network when a centralized service may suffer from poor connectivity and long latency. Different traditional cloud, mobile smaller deployed closer users, typically attached cellular base station or wireless point. Due relatively small coverage area each point, user moves across areas covered by different stations points are micro-clouds, issues...

Federated learning (FL) is a useful tool in distributed machine that utilizes users' local datasets privacy-preserving manner. When deploying FL constrained wireless environment; however, training models time-efficient manner can be challenging task due to intermittent connectivity of devices, heterogeneous connection quality, and non-i.i.d. data. In this paper, we provide novel convergence analysis non-convex loss functions using on both i.i.d. with arbitrary device selection probabilities...

"Edge-clouds," which are small servers located close to mobile users, have the potential greatly reduce delay and backhaul traffic of applications by moving cloud services closer users at edge. Due their limited storage capacity, proper configurations edge-clouds a significant impact on performance. This paper proposes tractable online algorithm that configures dynamically solely based past system history without any assumptions arrival patterns applications. We evaluate competitive ratio,...

Edge servers, which are small servers located close to mobile users, have the potential greatly reduce delay and backhaul traffic of Internet applications by moving cloud services edge network. Due limited capacity dynamic request arrival, proper service caching at is essential guarantee good performance. This paper proposes a tractable online algorithm called retrospective download with least-requested deletion that caches dynamically without any assumptions on arrival patterns...

This paper considers a Mobile-Edge Computing (MEC) enabled wireless network where the MEC-enabled Base Station (MBSs) can host application services and execute computation tasks corresponding to these when they are offloaded from resource-constrained mobile users. We aim at addressing joint problem of service caching—the provisioning their related libraries/database MBSs—and task-offloading assignment in densely-deployed each user exploit degrees freedom offloading different portions its...

In this paper, we consider user location privacy in mobile edge clouds (MECs). MECs are small deployed at the network to offer cloud services close users, and many solutions have been proposed maximize service locality by migrating follow their users. Co-location of a his service, however, implies that cyber eavesdropper observing migrations between can localize up one MEC coverage area, which be fairly (e.g., femtocell). We using chaff defend against such an eavesdropper, with focus on...

Most of the work on opinion dynamics models focuses case two or three types. We consider an arbitrary number opinions in mean field naming game model which it is assumed population infinite and all individuals are neighbors. A particular challenge that variables, corresponds to possible sets opinions, grows exponentially with opinions. present a method for generating dynamical equations general k calculate steady states important special cases arbitrarily high dimension: there exist zealots...

With the proliferation of fairly powerful mobile devices and ubiquitous wireless technology, we see a transformation from traditional ad hoc networks (MANETs) into new era service-oriented MANETs wherein node can provide receive services. Requested services must be decomposed more abstract then bound; formulate this as multi-objective optimization (MOO) problem to minimize service cost, while maximizing quality information in user receives. The MOO is an SP-to-service assignment problem. We...

In the future, video-enabled camera will be most pervasive type of sensor in Internet Things. Such cameras enable continuous surveillance through heterogeneous networks consisting fixed systems as well on mobile devices. The challenge these is to efficient video analytics: ability process videos cheaply and quickly searching for specific events or sequences events. this paper, we discuss design implementation Kestrel, a analytics system that tracks path vehicles across network. feeds are...