S-CMA: Sporadic Cloud-based Mobile Augmentation supported by an Ad-hoc Cluster of Moving Handheld Devices and a Virtualization Layer Esteban F. Ord´ o˜ nez-Morales * , Yolanda Blanco-Fern´ andez † , Jack F. Bravo-Torres ‡ , Mart´ ın L´ opez-Nores † , V´ ıctor Sai´ ans-V´ azquez † and Jos´ e J. Pazos-Arias † * Centro de Investigaci´ on, Desarrollo e Innovaci´ on en Ingenier´ ıas, Universidad Polit´ ecnica Salesiana, Cuenca, Ecuador Email: eordonez@ups.edu.ec † AtlantTIC Research Center, Department of Telematics Engineering, University of Vigo, Vigo, Spain Email: {yolanda,mlnores,vsaians,jose}@det.uvigo.es ‡ ´ Area de Ciencias Exactas, Universidad Polit´ ecnica Salesiana, Cuenca, Ecuador Email: jbravo@ups.edu.ec Abstract—Cloud-based Mobile Augmentation (CMA) is aimed to increase the computing capabilities of handheld devices for enabling the execution of resource-intensive mobile applications. In order to avoid latencies and excessive bandwidth consumption, the most recent CMA approaches borrow the augmentation resources from proximate clusters of handheld devices which do remain immobile, being their mobility an open challenge in literature. This paper proposes a new CMA solution where the resources are lent by an ad-hoc cluster of moving handheld devices. The complexity derived from the cluster mobility is tackled by a virtualization layer which handles stationary virtual nodes that are emulated by the users’ mobile devices. This way our CMA approach takes on sharing and allocating the resources available in the mobile ad-hoc cluster, ignoring the mobility of its devices. This contributes to alleviate limitations of existing CMA solutions that degrade significantly the experience of mobile users. I. MOTIVATION AND ANTECEDENTS The technological capabilities of current handheld devices 1 are propitious to deploy the so-called Sporadic Ad-hoc Net- works (hereafter SANs) among a group of moving users who are near from each other during a period of time [1]. These networks are appropriate to deliver context-aware tailor-made applications aimed at promoting interactions among nearby strangers with potentially similar interests and needs. Actually, the applications to be deployed over a SAN cover a wide spectrum, ranging from the orchestation of activities bounded to an event where a group of like-minded users happen to meet (e.g. in a museum, stadium, concert hall...), to the provision of applications for intervehicular communication and even refinements in the context of the smart cities through the planification of people mobility and urban games. In this paper, we focus in a particular domain in order to illustrate clearly how our approach works. We assume a vehicular communication environment where diverse functionalities can be delivered over a SAN, including optimization of traffic flows, chats among drivers, or other more sophisticated options like proactive organization of ride-sharing opportunities or selective distribution of personalized advertising in nearby places. 1 In this paper we use device and terminal as synonymous terms. Running this kind of computing-intensive mobile applica- tions requires to increase, enhance and optimize computing capabilities of resource-constraint mobile devices, by resorting to Mobile Cloud Computing (MCC) principles and infrastruc- tures [2]. Recent solutions are based on the so-called Cloud- based Mobile Augmentation (CMA) approaches, which consist of leveraging cloud-based resources (both distant and nearby clouds) to meet the requirements of mobile users, by offloading (part of) the code of the applications and the data required for their execution to a remote server. According to the taxonomy given in [3], cloud-based infrastructures adopted in traditional CMA solutions can be categorized into three main types –distant immobile clouds, proximate immobile clouds and proximate mobile clouds– that can be even combined in hybrid approaches. • The distant immobile clouds are supported by public and private clouds that comprise a large number of stationary remote servers that are typically located far from mobile devices. Long distances while passing data between handheld devices and cloud lead to high latencies, which significantly degrade the quality of the user experience when interacting with mobile applications (especially in low-bandwidth, intermittent networks [4]). • In order to tackle problems related to bandwidth and latency, some CMA solutions adopt proximate immo- bile clouds (typically named cloudlets) that involve stationary computers located in public places near the mobile devices (e.g., shopping malls, cinemas, airports...). Besides the fact that the cloudlet resources are frequently underutilized, these approaches limit significantly the users’ mobility because resource- intensive mobile applications can be just executed in the area around the cloudlet [5]. • The most recent CMA approaches rely on proximate mobile clouds where nearby handheld devices lend their resources to other mobile terminals to perform intensive computations in distributed manner [6, 7, 8]. 152