arXiv subjects

Alberto Leon-Garcia

Publications and source records attributed to Alberto Leon-Garcia.

25 records · Page 2Linked to original sources

arXiv2019

Artificial Intelligence as a Services (AI-aaS) on Software-Defined Infrastructure

This paper investigates a paradigm for offering artificial intelligence as a service (AI-aaS) on software-defined infrastructures (SDIs). The increasing complexity of networking and computing infrastructures is already driving the introduction of automation in networking and cloud computing management systems. Here we consider how these automation mechanisms can be leveraged to offer AI-aaS. Use cases for AI-aaS are easily found in addressing smart applications in sectors such as transportation, manufacturing, energy, water, air quality, and emissions. We propose an architectural scheme based on SDIs where each AI-aaS application is comprised of a monitoring, analysis, policy, execution plus knowledge (MAPE-K) loop (MKL). Each application is composed as one or more specific service chains embedded in SDI, some of which will include a Machine Learning (ML) pipeline. Our model includes a new training plane and an AI-aaS plane to deal with the model-development and operational phases of AI applications. We also consider the role of an ML/MKL sandbox in ensuring coherency and consistency in the operation of multiple parallel MKL loops. We present experimental measurement results for three AI-aaS applications deployed on the SAVI testbed: 1. Compressing monitored data in SDI using autoencoders; 2. Traffic monitoring to allocate CPUs resources to VNFs; and 3. Highway segment classification in smart transportation.

cs.LG↗

arXiv2018

Autonomous Mobility and Energy Service Management in Future Smart Cities: An Overview

With the rise of transportation electrification, autonomous driving and shared mobility in urban mobility systems, and increasing penetrations of distributed energy resources and autonomous demand-side management techniques in energy systems, tremendous opportunities, as well as challenges, are emerging in the forging of a sustainable and converged urban mobility and energy future. This paper is motivated by these disruptive transformations and gives an overview of managing autonomous mobility and energy services in future smart cities. First, we propose a three-layer architecture for the convergence of future mobility and energy systems. For each layer, we give a brief overview of the disruptive transformations that directly contribute to the rise of autonomous mobility-on-demand (AMoD) systems. Second, we propose the concept of autonomous flexibility-on-demand (AFoD), as an energy service platform built directly on existing infrastructures of AMoD systems. In the vision of AFoD, autonomous electric vehicles provide charging flexibilities as a service on demand in energy systems. Third, we analyze and compare AMoD and AFoD, and we identify four key decisions that, if appropriately coordinated, will create a synergy between AMoD and AFoD. Finally, we discuss key challenges towards the success of AMoD and AFoD in future smart cities and present some key research directions regarding the system-wide coordination between AMoD and AFoD.

cs.CY↗

arXiv2017

Elascale: Autoscaling and Monitoring as a Service

Auto-scalability has become an evident feature for cloud software systems including but not limited to big data and IoT applications. Cloud application providers now are in full control over their applications' microservices and macroservices; virtual machines and containers can be provisioned or deprovisioned on demand at runtime. Elascale strives to adjust both micro/macro resources with respect to workload and changes in the internal state of the whole application stack. Elascale leverages Elasticsearch stack for collection, analysis and storage of performance metrics. Elascale then uses its default scaling engine to elastically adapt the managed application. Extendibility is guaranteed through provider, schema, plug-in and policy elements in the Elascale by which flexible scalability algorithms, including both reactive and proactive techniques, can be designed and implemented for various technologies, infrastructures and software stacks. In this paper, we present the architecture and initial implementation of Elascale; an instance will be leveraged to add auto-scalability to a generic IoT application. Due to zero dependency to the target software system, Elascale can be leveraged to provide auto-scalability and monitoring as-a-service for any type of cloud software system.

cs.DC↗

arXiv2017

A Load-Balanced Call Admission Controller for IMS Cloud Computing

Network functions virtualization provides opportunities to design, deploy, and manage networking services. It utilizes cloud computing virtualization services that run on high-volume servers, switches, and storage hardware to virtualize network functions. Virtualization techniques can be used in IP multimedia subsystem (IMS) cloud computing to develop different networking functions (e.g., load balancing and call admission control). IMS network signaling happens through session initiation protocol (SIP). An open issue is the control of overload that occurs when an SIP server lacks sufficient CPU and memory resources to process all messages. This paper proposes a virtual load balanced call admission controller (VLB-CAC) for the cloud-hosted SIP servers. VLB-CAC determines the optimal call admission rates and signaling paths for admitted calls along with the optimal allocation of CPU and memory resources of the SIP servers. This optimal solution is derived through a new linear programming model. This model requires some critical information of SIP servers as input. Further, VLB-CAC is equipped with an autoscaler to overcome resource limitations. The proposed scheme is implemented in smart applications on virtual infrastructure (SAVI) which serves as a virtual testbed. An assessment of the numerical and experimental results demonstrates the efficiency of the proposed work.

cs.NI↗

arXiv2017

OpenSIP: Toward Software-Defined SIP Networking

VoIP is becoming a low-priced and efficient replacement for PSTN in communication industries. With a widely growing adoption rate, SIP is an application layer signaling protocol, standardized by the IETF, for creating, modifying, and terminating VoIP sessions. Generally speaking, SIP routes a call request to its destination by using SIP proxies. With the increasing use of SIP, traditional configurations pose certain drawbacks, such as ineffective routing, un-optimized management of proxy resources (including CPU and memory), and overload conditions. This paper presents OpenSIP to upgrade the SIP network framework with emerging technologies, such as SDN and NFV. SDN provides for management that decouples the data and control planes along with a software-based centralized control that results in effective routing and resource management. Moreover, NFV assists SDN by virtualizing various network devices and functions. However, current SDN elements limit the inspected fields to layer 2-4 headers, whereas SIP routing information resides in the layer-7 header. A benefit of OpenSIP is that it enforces policies on SIP networking that are agnostic to higher layers with the aid of a Deep Packet Inspection (DPI) engine. Among the benefits of OpenSIP is programmability, cost reduction, unified management, routing, as well as efficient load balancing. The present study implements OpenSIP on a real testbed which includes Open vSwitch and the Floodlight controller. The results show that the proposed architecture has a low overhead and satisfactory performance and, in addition, can take advantage of a flexible scale-out design during application deployment.

cs.NI↗

arXiv2016

Leveraging Synergy of 5G SDWN and Multi-Layer Resource Management for Network Optimization

Fifth-generation (5G) cellular wireless networks are envisioned to predispose service-oriented, flexible, and spectrum/energy-efficient edge-to-core infrastructure, aiming to offer diverse applications. Convergence of software-defined networking (SDN), software-defined radio (SDR) compatible with multiple radio access technologies (RATs), and virtualization on the concept of 5G software-defined wireless networking (5G-SDWN) is a promising approach to provide such a dynamic network. The principal technique behind the 5G-SDWN framework is the separation of the control and data planes, from the deep core entities to edge wireless access points (APs). This separation allows the abstraction of resources as transmission parameters of each user over the 5G-SDWN. In this user-centric and service-oriented environment, resource management plays a critical role to achieve efficiency and reliability. However, it is natural to wonder if 5G-SDWN can be leveraged to enable converged multi-layer resource management over the portfolio of resources, and reciprocally, if CML resource management can effectively provide performance enhancement and reliability for 5G-SDWN. We believe that replying to these questions and investigating this mutual synergy are not trivial, but multidimensional and complex for 5G-SDWN, which consists of different technologies and also inherits legacy generations of wireless networks. In this paper, we propose a flexible protocol structure based on three mentioned pillars for 5G-SDWN, which can handle all the required functionalities in a more crosslayer manner. Based on this, we demonstrate how the general framework of CML resource management can control the end user quality of experience. For two scenarios of 5G-SDWN, we investigate the effects of joint user-association and resource allocation via CML resource management to improve performance in a virtualized network.

cs.NI↗

arXiv2015

Virtualization of Multi-Cell 802.11 Networks: Association and Airtime Control

This paper investigates the virtualization and optimization of a multi-cell WLAN. We consider the station (STA)-access point (AP) association and airtime control for virtualized 802.11 networks to provide service customization and fairness across multiple internet service providers (ISPs) sharing the common physical infrastructure and network capacity. More specifically, an optimization problem is formulated on the STAs transmission probabilities to maximize the overall network throughput, while providing airtime usage guarantees for the ISPs. Subsequently, an algorithm to reach the optimal solution is developed by applying monomial approximation and geometric programming iteratively. Based on the proposed three-dimensional Markov-chain model of the enhanced distributed channel access (EDCA) protocol, the detailed implementation of the optimal transmission probability is also discussed. The accuracy of the proposed Markov-chain model and the performance of the developed association and airtime control scheme are evaluated through numerical results.

cs.NI↗