Cloud Computing and IoT: Benefits, Challenges, and Application Scenarios, Apuntes de Informática

¡Descarga Cloud Computing and IoT: Benefits, Challenges, and Application Scenarios y más Apuntes en PDF de Informática solo en Docsity! Integration of Cloud Computing and Internet of Things: a Survey Alessio Botta, Walter de Donato, Valerio Persico, Antonio Pescapé University of Napoli Federico 11 (Italy) NM2 SRL, Italy La.botta, walter.dedonato, valerio.persico, pescape) Qunina.it Abstract Cloud computing and Internet of Things (loT) are two very different tech- nologies that are both already part of our life. Their adoption and use is expected to be more and more pervasive, making them important compo- nents of the Future Internet. A novel paradigm where Cloud and loT are merged together is foreseen as disruptive and as an enabler of a large number of application scenarios. In this paper, we focus our attention on the integration of Cloud and loT, which is what we call the CloudloT paradigm. Many works in litera- ture have surveyed Cloud and loT separately and, more precisely, their main properties, features, underlying technologies, and open issues. However, to the best of our knowledge, these works lack a detailed analysis of the new CloudloT paradigm, which involves completely new applications, challenges, and research issues. To bridge this gap, in this paper we provide a literature survey on the integration of Cloud and loT. Starting by analyzing the basics of both loT and Cloud Computing, we discuss their complementarity, detail- ing what is currently driving to their integration. Thanks to the adoption of the CloudloT paradigm a number of applications are gaining momentum: we provide an up-to-date picture of CloudloT applications in literature, with a focus on their specific research challenges. These challenges are then ana- lyzed in details to show where the main body of research is currently heading. We also discuss what is already available in terms of platforms — both propri- etary and open source — and projects implementing the CloudloT paradigm. Finally, we identify open issues and future directions in this field, which we expect to play a leading role in the landscape of the Future Internet. Preprint submitted to Journal of Future Generation Computer Systems September 18, 2015 Keywords: Cloud Computing, Internet of Things, Ubiquitous Networks, Cloud of Things, Pervasive Applications, Smart City, Smart Applications. 1. INTRODUCTION AND MOTIVATION The Internet of Things (loT) paradigm is based on intelligent and self configuring nodes (things) interconnected in a dynamic and global network infrastructure. It represents one of the most disruptive technologies, enabling ubiquitous and pervasive computing scenarios. loT is generally characterized by real world small things, widely distributed, with limited storage and pro- cessing capacity, which involve concerns regarding reliability, performance, security, and privacy. On the other hand, Cloud computing has virtually un- limited capabilities in terms of storage and processing power, is a much more mature technology, and has most of the loT issues at least partially solved. Thus, a novel IT paradigm in which Cloud and loT are two complementary technologies merged together is expected to disrupt both current and Future Internet (132, 25]. We call this new paradigm CloudIoT. Reviewing the rich and articulate state of the art in this field, we found that both topics gained popularity in the last few years (Fig. la), and the number of papers dealing with Cloud and loT separately shows an increasing trend since 2008 (Fig. 1b)!. In this paper we review the literature focusing on the integration of Cloud and loT, a really promising topic for both research and industry, witnessed by the more recent and rapidly increasing trend dealing with Cloud and loT together (Fig. 1c). Inspired by well known indications in literature [74], we adopt the research methodology schematically depicted in Fig. 2. By analyzing a large number of papers mainly published between 2008 and 2014 in selected venues, (A) we derive a temporal characterization of the literature — aiming at showing in a qualitative way the temporal behavior of the research and the common interest about the CloudloT paradigm (see Fig. 1) — and build the basis for the following steps. The characterization of the literature is reported in this section and is supported by Fig. 1. (B) We introduce a short background on both Cloud and loT to provide the readers with the necessary basics and to tackle the integration of Cloud and loT (Sec. 2). (C) We present a detailed ¡Data have been obtained from Google Trends (https : //www.google.com/trends/) and Scholar (https: //scholar.google.com/) web facilities. sE A f 1 pes “En A y SÓ Figure 2: The research methodology adopted in this work. nologies with potential impact on US interests towards 2025 [60]. Indeed, in 2011 the number of interconnected devices overtook the number of peo- ple [53]. In 2012, the number of interconnected devices was estimated to be 9 billion, and it was expected to reach the value of 24 billions by 2020. Such numbers suggest that loT will be one of the main sources of big data [37]. In the following we describe a few important aspects related to loT. RFID. In loT scenario, a key role is played by Radio-Frequency IDentifi- cation (RFID) systems, composed of one or more readers and several tags. These technologies help in automatic identification of anything they are at- tached to, and allow objects to be assigned unique digital identities, to be integrated into a network, and to be associated with digital information and services [12]. In a typical usage scenario, readers trigger the tag transmis- sion by generating an appropriate signal, querying for possible presence of objects uniquely identified by tags. RFID tags are usually passive (they do not need on-board power supply), but there are also tags powered from Figure 3: loT paradigm: an overall view (source: [42)). batteries [12, 124]. (Wireless) Sensor Networks. Another key component in loT environ- ments is represented by sensor networks. For example, they can cooperate with RFID systems to better track the status of things, getting information about position, movement, temperature, etc. Sensor networks are typically composed of a potentially high number of sensing nodes, communicating in a wireless multi-hop fashion. Special nodes (sinks) are usually employed to gather results. Wireless sensor networks (WSNs) may provide various useful data and are being utilized in several areas like healthcare, govern- ment and environmental services (natural disaster relief), defense (military target tracking and surveillance), hazardous environment exploration, seis- mic sensing, etc. [7]. However, sensor networks have to face many issues regarding their communications (short communication range, security and privacy, reliability, mobility, etc.) and resources (power considerations, stor- age capacity, processing capabilities, bandwidth availability, etc.). Besides, WSN has its own resource and design constraints (that are application- and environment- specific) and that heavily depend on the size of the monitoring environment [7]. The scientific community deeply addressed several issues re- lated to sensor networks at different layers (e.g., energy efficiency, reliability, robustness, scalability, etc.) [12]. Addressing. Thanks to wireless technologies such as RFID and Wi-Fi, loT' paradigm is transforming the Internet into a fully integrated Future Inter- net [124]. While Internet evolution led to an unprecedented intercomnection of people, current trend is leading to the interconnection of objects, to create a smart environment [53]. In this context, the ability to uniquely identify things is critical for the success of loT since this allows to uniquely address a huge number of devices and control them through the Internet. Uniqueness, reliability, persistence, and scalability represent critical features related to the creation of a unique addressing schema [53]. Unique identification issues may be addressed by IPv4 to an extent (usually a group of cohabiting sensor devices can be identified geographically, but not individually). IPv6, with its Internet Mobility attributes, can mitigate some of the device identification problems and is expected to play an important role in this field. Middleware. Due to the heterogeneity of the participating objects, to their limited storage and processing capabilities and to the huge variety of appli- cations involved, a key role is played by the middleware between the things and the application layer, whose main goal is the abstraction of the func- tionalities and communication capabilities of the devices. The middleware can be divided in a set of layers (see Fig. 3): Object Abstraction, Service Management, Service Composition, and Application [42]. 2.2. Cloud The essential aspects of Cloud computing have been reported in the definition provided by the National Institute of Standard and Technologies (NIST) [92]: “Cloud computing is a model for enabling ubiquitous, conve- nient, on-demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage, applications, and services) that can be rapidly provisioned and released with minimal management effort or service provider interaction.” Even though the main idea behind Cloud com- puting was not new, the term started to gain popularity after that Google's CEO Eric Shmidt used it in 2006 [130], and over the last few years the appearance of Cloud computing has hugely impacted IT industry. The avail- ability of virtually unlimited storage [113] and processing capabilities at low cost enabled the realization of a new computing model, in which virtualized resources can be leased in an on-demand fashion, being provided as general utilities. Large companies (like Amazon, Google, Facebook, etc.) widely adopted this paradigm for delivering services over the Internet, gaining both economical and technical benefits. Cloud Computing is a disruptive technology with profound implications for the delivery of Internet services as well as for the IT sector as a whole. 7 Table 1: Complementary aspects of Cloud and loT. ToT Cloud Displacement pervasivo contralized Reachability limited ubiquitous Components | real world things virtual resourcos Computational capabilities limited virtually unfimite Storage | — limited or none virtually unhimited Role of the Internet | point of convergen: means for delivering services Big data source means to manage and are proposing their integration, generally to obtain benefits in specific application scenarios (8, 4, 51]. In general, loT can benefit from the virtually unlimited capabilities and resources of Cloud to compensate its technological constraints (e.g., storage, processing, communication). To cite a few examples, Cloud can offer an ef- fective solution for loT service management and composition as well as for implementing applications and services that exploit the things or the data produced by them [83]. On the other hand, Cloud can benefit from loT by extending its scope to deal with real world things in a more distributed and dynamic manner, and for delivering new services in a large number of real life scenarios. In many cases, Cloud can provide the intermediate layer between the things and the applications, hiding all the complexity and functionali- ties necessary to implement the latter. This will impact future application development, where information gathering, processing, and transmission will generate new challenges, especially in a multi-cloud environment [41]. In this section we discuss the main CloudloT drivers, i.e., the motivations driving toward the integration of Cloud and loT. Most of the papers in literature are actually seeing Cloud as the missing piece in the integrated scenario, i.e. they believe that Cloud fills some gaps of loT (e.g. the limited storage). A few others, instead, see loT filling gaps of Cloud (mainly the limited scope). In this paper we consider both as CloudloT drivers and we start our discussion from the first ones. Most of these drivers fall in three categories that are communication, storage, and computation, while a few others are more basic and have impli- cations in all such categories, they are transversal. In the following, we start discussing such transversal drivers, and then detail the ones related to communication, storage, and processing. Being loT characterized by a very high heterogeneity of devic nologies, and protocols, it lacks different important properties such as scala- tech- 10 bility, interoperability, flexibility, reliability, efficiency, availability, and secu- rity. Since Cloud has proved to provide them [47, 34, 115], we identify them as some of the main transversal CloudloT drivers. Two other transversal drivers are the ease of use and the reduced cost obtained by both users and providers of applications and services [34]. Indeed, Cloud facilitates the flow between loT data collection and data processing, and enables rapid setup and integration of new things, while maintaining low costs for deployment and for complex data processing [110]. As a consequence, analyses of unprece- dented complexity [34, 101) are possible, and data-driven decision making and prediction algorithms can be employed at low cost, providing means for increasing revenues and reduced risks [129]. Communication. Data and application sharing are two important CloudloT drivers falling in the communication category. Thanks to the CloudloT paradigm, personalized ubiquitous applications can be delivered through the loT, while automation can be applied to both data collection and distribution at low cost. Cloud offers an effective and cheap solution to connect, track, and manage any thing from anywhere at any time by using customized por- tals and built-in apps [110]. The availability of high speed networks enables effective monitoring and control of remote things [110, 47, 101], their coordi- nation [47, 101, 115], their communications [47], and real-time access to the produced data [110]. It is worth mentioning that although Cloud can significantly improve and simplify loT communication, it can still represent a bottleneck in some scenarios: indeed, over the last 20 years data storage density and processor power increased of a factor of 10% and 10% respectively, while broadband capacity increased only of 10* [68]. As a consequence, practical limitations can arise when trying to transfer huge amounts of data from the edge of the Internet onto Cloud. Storage. loT involves by definition a large amount of information sources (i.e., the things), which produce a huge amount of non-structured or semi- structured data [41], which also have the three characteristics typical of Big Data [134]: volume (i.e., data size), variety (i.e., data types), and velocity (i.e., data generation frequency). Large-scale and long-lived storage, possible thanks to the virtually unlimited, low-cost, and on-demand storage capacity provided by Cloud, represents an important CloudloT driver. Cloud is the most convenient and cost effective solution to deal with data produced by loT [110] and, in this respect, it generates new opportunities for data aggre- 11 gation [47], integration [129], and sharing with third parties [129]. Once into Cloud, data can be treated as homogeneous through well-defined APIs [47], can be protected by applying top-level security [34], and can be directly accessed and visualized from any place [110]. Computation. loT devices have limited processing and energy resources that do not allow complex, on-site data processing. Collected data is usu- ally transmitted to more powerful nodes where aggregation and processing is possible, but scalability is challenging to achieve without a proper infras- tructure. Cloud offers virtually unlimited processing capabilities and an on- demand usage model. This represents another important CloudloT driver: loT processing needs can be properly satisfied for performing real-time data analysis (on-the-fly) [110, 34], for implementing scalable, real-time, collab- orative, sensor-centric applications [47], for managing complex events [110], and for supporting task oflloading for energy saving [125]. Scope. As the things add capabilities, and more people and new types of information are connected, users spread across the world quickly enter the Internet of Everything (1oE) [43, 2], a network of networks where billions of connections create unprecedented opportunities as well as new risks. The adoption of the CloudloT paradigm enables new smart services and appli- cations based on the extension of Cloud through the things (110, 115] which enable the cloud to deal with a number of new, real-life scenarios, giving birth to the Things as a Service paradigm [27, 94, 36]. This is another important driver for CloudloT. The literature shows how a number of new paradigms emerging from the integration of Cloud and loT and related to this particular driver. They are summarized in Tab. 2. Since no standard has been clearly defined, there is no sharp distinction among the proposed acronyms, which in some cases appear to collide. Vehicular Cloud is another important new paradigm emerging in this area [57]. In conclusion, several motivations are driving the integration of Cloud and loT. Some of them are actually related with specific application scenarios. Section 4 analyzes these application scenarios in details, revealing the main challenges associated with each of them. 12 cation, also on the go [97]. Cloud allows to face common challenges of this application scenario such as: security, privacy, and reliability, by enhanc- ing medical data security and service availability and redundancy [77, 6]. Thanks to the efficient management of sensor data it is possible to provide assisted-living services in real-time [44]. Moreover, Cloud adoption enables the execution (in the Cloud) of secure multimedia-based health services, over- coming the issue of running heavy multimedia and security algorithms on devices with limited computational capacity and small batteries [97], and it provides a flexible storage and processing infrastructure to perform both online and offline analyses of data streams generated in healthcare Body Sen- sor Networks (BSNs)?. Thanks to the use of the CloudloT paradigm BSNs can be deployed in a community of people and can generate large amounts of contextual data that are stored, processed, and analyzed in a scalable fashion [45]. In the healthcare domain, common challenges are related to the lack of trust in data security and privacy by users (exposure to hacker attacks, viola- tion of medical data confidentiality, data lock-in and loss of governance, privi- lege abuse), performance unpredictability (resource exhaustion, data transfer bottlenecks, impact on real-time services, streaming QoS), legal issues (con- tract law, intellectual property rights, data jurisdiction) and are still object of investigation [77, 38, 97]. The lack of specific research related to the adoption of these technologies in the context of mission critical systems, of deepened reliability analyses and the limited number of case studies are often defined as the major obstacles [77, 44, 45]. Smart Cities and Communities. CloudloT leads to the generation of services that interact with the surrounding environment, thus creating new opportunities for contextualization and geo-awareness. Sustainable devel- opment of urban areas is a challenge of key importance and requires new, efficient, and user-friendly technologies and services. The challenge is to harness the collaborative power of ICT networks (networks of people, of knowledge, of sensors) to create collective and individual awareness about the multiple sustainability threats which our society is facing nowadays at social, environmental, and political levels. The resulting collective intelli- 2BSNs have been recently introduced for the remote monitoring of human activities in > domains but also in other application domains such as emergency manage- and behavior surveillance. gence will lead to better informed decision-making processes and empower citizens, through participation and interaction, to adopt more sustainable in- dividual and collective behaviours and lifestyles [58]. CloudloT can provide a common middleware for future-oriented smart-city services [13, 115, 28], acquiring information from different heterogeneous sensing infrastructures, accessing all kinds of geo-location and loT technologies (e.g., 3D representa- tions through RFID sensors and geo-tagging), and exposing information in a uniform way (e.g., through a dynamically annotated map). Frameworks typ- ically consist of a sensor platform (with APIs for sensing and actuating) and a Cloud platform which offers scalable and long-lived storage and processing resources for the automatic management and control of real-world sensing devices in a large-scale deployment. Crowdsourced and reputation-based frameworks also exist: authors of [72, 71] propose a framework implement- ing the Sensing as a Service paradigm in the context of smart cities and aimed at public safety. Authors of [9, 107] present an ecosystem for mobile crowdsensing applications which relies on the Cloud-based publish /subscribe middleware to acquire sensor data from mobile devices in a context-aware and energy-efficient manner. Authors of [121] focus on the burden on ap- plication developers and final users created by the need to deal with-large scale environments. Since loT scenario is highly fragmented, sensor virtual- ization can be employed to reduce the gap between existing heterogeneous technologies and their potential users, allowing them to interact with sensors at different layers [106]. A number of recently proposed solutions suggest to use Cloud architec- tures to enable the discovery, the connection, and the integration of sensors and actuators, thus creating platforms able to provision and support ubiqui- tous connectivity and real-time applications for smart cities [94, 105]. For instance, authors of [76] discuss a concept towards developing a smart city using an intelligent, energy efficient, public illamination system, which would also offer ubiquitous communication. Moreover, Cloud-based platforms help to make it easier for third parties to develop and provide loT plugins en- abling any device to be connected to the Cloud [13]. This type of advanced service model hides the complexity and the heterogeneity of the underly- ing infrastructure, while at the same time meeting complex requirements for Cloud, such as high reactivity and timeliness, scalability, security, easy- configurability, and flexibility [13, 115, 135]. Common challenges are related to security, reliability, scale, heterogene- ity and timeliness. Indeed, enabling the necessary resources, storage and 16 computing capabilities for large amounts of heterogeneous and personalized data (coming from distributed sources) in a transparent and secure manner and the development and the deployment of various middleware platforms in a such fragmented scenario (in which different loT ecosystems are not able to communicate between them) are not trivial tasks [115]. The involvement of multiple physical sensors in the scope of service delivery creates additional challenges associated with real-time interactions, which imposes a need for studying extensions over real-time operating systems for embedded devices, as well as how they could be supported in the scope of a Cloud environ- ment [115]. Moreover, the resulting system has to provide a rapid setup of deployed sensors and an easy integration of new sensors in the sensing environment [94]. The blending of loT resources into the Cloud introduces new resource management requirements, which are associated with the need to optimize not only processing, storage and I/O resources, but also sensor reading cy- cles, multi-sensor queries and shared access to expensive location-dependent loT resources [115]. Significant research on sensing, actuation, and loT is di- rected towards the efficient semantic annotation of sensor data [94]. Finally, while cities share common concerns — such as the need to effectively share in- formation within and between cities and the desire for enhanced cross-border protocols — they lack a common infrastructure and methodology for collab- orating, generating operational and regional fragmentation that currently prevent innovative synergies [13, 115]. Smart Home and Smart Metering. Home networks have been identified as the environment where users mainly act: CloudloT has large application in home environments, where the joint adoption of heterogeneous embedded devices and Cloud enables the automation of common in-house activities. In- deed, the merging of computing with physical things, enables the transforma- tion of everyday objects into information appliances which — interconnected through the Internet — can expose services through a web interface. Sev- eral smart-home applications proposed in literature involve (wireless) sensor networks and realize the connection of intelligent appliances to the Inter- net in order to remotely monitor their behavior (e.g., to monitor devices” power usage to improve power usage habits [26)) or remotely control them (e.g., to manage lighting, heating, and air conditioning [54]). In partic- ular, smart lighting recently attracted growing attention from the research community [127, 91]; lighting is responsible for 19% of global use of electri- 17 different types of Clouds, which also include temporary vehicular Clouds (i.e. formed by the vehicles representing the Cloud datacenters [18]) de- signed to expand the conventional Clouds in order to increase on-demand the whole Cloud computing, processing, and storage capabilities, by using under-utilized facilities of vehicles. Several challenges have been identified in literature related to this appli- cation scenario. The huge number of vehicles and their dynamically changing number make system scalability difficult to achieve [57]. Vehicles moving at various speeds frequently cause intermittent communication impacting per- formance, reliability, and QoS [18]. The lack of an established infrastructure makes it difficult to implement effective authentication and authorization mechanisms [18], with impacts on security and privacy provision [90]. The lack of global standards, experimental studies, and proper benchmarks on realistic ITS-Clouds afects interoperability [57]. Smart Energy and Smart Grid. loT and Cloud can be effectively merged to provide intelligent management of energy distribution and consumption in both local and wide-area heterogeneous environments. The loT nodes typically involved in this kind of processes have sensing, processing, and networking capabilities, but limited resources. Hence, com- puting tasks can be properly demanded to the Cloud, where more complex and comprehensive decisions can be made. Cloud adoption leads to increase the reliability by providing sel£healing mechanisms and enables mutual op- eration and participation of the users, to achieve distributed generation, elec- tricity quality, and demand response [128]. Cloud computing makes possible to analyze and process vast amounts of data and information coming from. different sources distributed along wide area networks, for the purpose of implementation of intelligent control to objects. Several challenges should be adequately addressed to realize the full po- tential of such application. Large scale distributed sources raise issues about heterogeneity, data size and collection rate, latency dynamics, and cost of security enforcement [128]. Security and privacy concerns inherent in an information-rich smart grid evironment may be further exacerbated by de- ployment on Cloud and introduce challenges such as: the integration of data having diverse ownership, the aggregation of public and private data, or a longer and wider exposure to attacks [111]. Legal issues can derive from the distribution of data archival over different jurisdictions [111]. Finally, con- sumers should gain more confidence in sharing data to help improving and 20 optimizing services offered [111]. Smart Logistics. The adoption of CloudloT in logistics promotes a new service mode that is radically changing business paradigms [85]. It enables new interesting scenarios and allows the easy and automated management of flows of goods between the point of origin and the point of consumption, in order to meet specific requirements expressed in terms of time, cost or means of transport. Moreover, thanks to geo-tagging technologies, it enables to automatically track goods while in transit. Cloudlo'T' is proposed to help conventional logistics systems in evolving into advanced systems, capable of dealing automatically with complexity and changes. Ideed, logistics resources are heterogeneous (e.g., geographical distribution, morphological diversity, and sel£governing zone). These make resource sharing and management more complex. Hence, computer aided software tools supporting the adoption of loT can experience a bottleneck in dealing with complexity, dynamics, and uncertainties in their applica- tions in modern enterprises. The adoption of Cloud computing can help in overcoming the bottlenecks enabling complex decision-making systems where automated algorithms can be enforced to retrieve information for assembly planning [119]. By adopting a scalable and modularized architecture, Cloud helps to make the system robust, reliable, flexible, and easily expandable. In this context, important challenges are related to resource heterogene- ity, and solutions are investigated in terms of logistics virtualization and service selection [85]. The former is critical for resource sharing and dynamic allocation and provides flexibility in the use of resources. The latter allows service requesters and providers to agree on the attributes that govern the in- teraction and provides the selection of an appropriate web service that meets certain functional and non-functional criteria. Environmental Monitoring. The combined use of Cloud and loT can contribute to the deployment of a high speed information system between the entity in charge of monitoring wide-area environments and the sen- sors/actuators properly deployed in the area. Some applications can be re- lated to the continuous and long-term monitoring of water level (for lakes, streams, sewages), gas concentration in air (e.g., in laboratories, deposits), soil humidity and other characteristics, inclination for static structures (e.g., bridges, dams), position changes (e.g., land slides), lighting conditions (e.g., to detect intrusions in dark places), infrared radiation for fire or animal de- tection [78]. Other potential applications of this kind are: agriculture infor- 21 mation transmission and intelligent detection, intelligent cultivation control, food safety tracking, precision irrigation, forest identification, and tree track- ing [110]. A Cloud-based data access is able to bridge the latency-energy require- ments of low power communication segments and the ubiquitous and fast ac- cess to data for end users (either humans or loT applications) [78]. Moreover it allows to manage and process complex events, generated by the real-time data streamed by sensors. The main challenges in this field pertain to the potential massive-scale of the infrastructures. Specifically, environmental dynamism makes it difficult to provide computational resources that are sufficient to deal with changing environmental conditions. Moreover, challenges are also related to security, as threats can be found in information leak due to potential breaches caused by infected clients or communication channel vulnerabilities. Finally, research is still needed on the implementation and promotion of proper communication protocols (such as IPv6 for individually addressing the things), the setting of various loT standards for promoting interoperabil- ity and for scaling the cost of loT facilities, and the assessment of risks and uncertainties. In this section we have carefully surveyed and described a number of applications arising from the adoption of the CloudloT paradigm and for each of them we have pinpointed the related challenges. In Sec. 5 we provide a detailed discussion of these challenges. 5. CHALLENGES We have discussed how integrating Cloud and loT provides several bene- fits and fosters the birth or improvement of a number of interesting applica- tions. At the same time, we have seen that the complex CloudloT scenario imposes several challenges for each application that is currently receiving attention by the research community [41]. This section is devoted to the analysis of such challenges. In the following we first deal with the typical challenges raised by the application scenarios reported above. We then focus on other important recurring challenging topics strictly related to CloudloT. Security and Privacy. When critical loT applications move towards the Cloud, concerns arise due to the lack of, e.g., trust in the service provider, knowledge about service level agreements (SLAs), and knowledge about the 22 constrained environments a number of challenges related to device failure or not always reachable devices exists [57]. From the one hand, Cloud capabilities help to overcome some of these challenges (e.g., Cloud enhances the reliability of the devices by allowing to offload heavy tasks and thus to increase devices” battery duration or offering the possibility of building a modularized architecture) [128, 119]; on the other hand, it introduces uncertainties related to data center virtualization or resource exhaustion [110, 77]. The lack of reliability analyses and of the development of specific case studies exacerbate the challenge. Large Scale. CloudloT allows to design novel applications aimed at inte- grating and analyzing information coming from real-world (embedded) de- vices [105, 78, 19, 94, 110, 121]. Some of the depicted scenarios implicitly require the interaction with a very large number of these devices, usually distributed across wide-area environments. The large scale of the result- ing systems makes typical challenges harder to overcome (e.g., requirements about storage capacity and computational capability for further processing become arduous to be satisfied when facing long-lived data collected at high- rate). Moreover the distribution of the loT devices makes monitoring tasks harder since they have to face latency dynamics and connectivity issues. Legal and Social Aspects. These are two important challenges, partly related. Legal aspects are extremely important and actual in the current re- search for specific application scenarios. Think, for example, to a CloudloT' service based on user-provided data. In this case, on the one hand, the ser- vice provider has to conform to different international laws. And, on the other hand, users have to be provided with incentives in order to contribute to data collection. In more general terms, Social aspects are of interest for research and are currently considered an interesting challenge because, often, the investment into omnipresent Internet-capable devices is not reasonable in every scenario. It is more convenient to give the opportunity to users to participate in submitting data that represent a thing [11]. The authors of [32] have identified a set of issues to be addressed by any system which incorporates humans as a source of sensor data, in order to remain trusted by its users (such as integrating the qualitative observations generated by hu- mans with the machine-generated quantitative observations, or the need to characterize and manage data quality, reliability, reputation, and trustwor- thiness). Users could also be empowered with new building blocks and tools: accelerators, frameworks, and toolkits that enable the participation of users in loT as done in the Internet through Wikis and Blogs [33]. Such tools and techniques should enable researchers and design professionals to learn about user work, giving users an active role in technology design. To achieve this, these tools should allow users to easily experiment various design possibilities in a cost-effective way. Related to this challenge, researchers are trying to provide adequate tools for implementing a cooperative prototyping approach, where users and designers explore together applications and their relations. Besides the challenges reported above, other important aspects are cur- rently of large interest for the research community. They partly intersect with the challenges reported above, but they require a separate discussion as they involve a large body of research on their own. Big Data. With an estimated number of 50 billion devices that will be networked by 2020, specific attention must be paid to transportation, stor- age, access, and processing of the huge amount of data they will produce. Thanks to the recent development in technologies, loT will be one of the main sources of big data, and Cloud will enable to store it for long time and to perform complex analyses on it. The ubiquity of mobile devices and sen- sor pervasiveness, indeed call for scalable computing platforms (every day 2.5 quintillion bytes of data are created) [37]. Handling this data conveniently is a critical challenge, as the overall application performance is highly depen- dent on the properties of the data management service [37]. For instance, cloud-based methods for Big Data summarization based on the extraction of semantic feature are actually under investigation [69]. Hence, following the NoSQL movement, both proprietary and open source solutions adopt alternative database technologies for big data [31]: time-series, key-value, document store, wide column stores, and graph databases. Unfortunately, no perfect data management solution exists for the Cloud to manage big data [129]. Moreover, data integrity is an important factor, not only for its impact on the qualities of service, but also for security and privacy related aspects expecially on outsourced data[87]. Sensor Networks. Sensor networks have been defined as the major en- abler of loT [129] and as one of the five technologies that will shape the world, offering the ability to measure, infer, and understand environmental indicators, from delicate ecologies and natural resources to urban environ- ments [53]. Recent technological advances have made efficient, low-cost, and low-power miniaturized devices available for use in large-scale, remote sens- 26 ing applications [5]. Moreover, smartphones, even though limited by power consumption and reliability, come with a variety of sensors (GPS, accelerom- eter, digital compass, microphone, and camera), enabling a wide range of mobile applications in different domains of IoT. In this context, the timely processing of huge and streaming sensor data, subject to energy and network constraints and uncertainties, has been identified as the main challenge [131]. Cloud provides new opportunities in aggregating sensor data and exploiting the aggregates for larger coverage and relevancy, but at the same time af fects privacy and security [131]. Furthermore, being lack of mobility a typical aspect of common loT devices, the mobility of sensors introduced by smart- phones as well as wearable electronics represents a new challenge [103]. Monitoring. As largely documented in the literature, monitoring is an essential activity in Cloud environments for capacity planning, for managing resources, SLAs, performance and security, and for troubleshooting [3]. As a consequence, CloudloT inherits the same monitoring requirements from Cloud, but the related challenges are further affected by volume, variety, and velocity characteristics of loT. Fog Computing. Fog computing is an extension of classic Cloud computing to the edge of the network (as fog is a cloud close to the ground). It has been designed to support loT applications characterized by latency constraints and requirement for mobility and geo-distribution [20, 133, 1]. Even though computing, storage, and networking are resources of both the Cloud and the Fog, the latter has specific characteristics: edge location and location aware- ness implying low latency; geographical distribution and a very large number of nodes in contrast to centralized Cloud; support for mobility (through wire- less access) and real-time interaction (instead of batch processes); support for interplay with the Cloud. Authors of [89] proposed an analysis showing how building Fog computing projects is challenging. Indeed, the adoption of Fog-based approaches requires various specific algorithms and methodologies dealing with reliability of the networks of smart devices, and operating under specific conditions that ask for fault tolerant techniques. 6. PLATFORMS, SERVICES, AND RESEARCH PROJECTS In this section we describe open source and proprietary platforms for implementing the new vision and paradigms of Sec. 3 and the new CloudloT' applications of Sec. 4. Also, we survey research projects focused on the 27 cheap products that have full TCP/IP stack implemented and HTTP server on-board, which allows to interact with them using simple RESTful APIs. Finally, there are also several services (es. Xively [122], Open.Sen.se [99], ThingSpeak [118], CloudPlugs [29], Carriots [22]) that allow to collect data from things and to store these data on the Cloud offered by the service provider. These services typically provide an API and different example ap- plications to use the data collected by the things, which range from specific, proprietary things to open, and widely distributed ones (e.g. Arduino). This is the most common market trend is this area, allowing service providers to offer free subscriptions and make business out of data provided by the users. Starting from these services, companies have created toolkits for their inte- gration in CloudloT frameworks. For example, NetLab [63] is a toolkit for interaction among physical and digital objects (e.g. controlling video movies through arduino). NetLab has created two widgets called CouldIn and Cloud- Out that allow to interact with several CloudloT services. In particular, they allow to periodically send data from things to these services or to periodically retrieve data from these services. Compliant services include Xively (formerly COSM and Pachube), Open.Sen.se, and ThingSpeak. Hardware producers have also started to launch Cloud services where clients can upload their data. For example, at Synapse they created a component of their operating system (SNAP) that allow to send data to private and public Cloud and to manage the related tasks (operation, administration, maintenance, and provisioning) [116]. Recently Intel has also launched an initiative [59] that provides a software library for Galileo/Edison platforms (compatible with arduino) and a private Cloud where data can be stored by things based on Galileo /Edison platforms and accessed by applications though a public APL The sources of software as well as the designs of the hardware are released to the public (i.e. open source and open hardware). 6.2. Research Projects ClouT [28], which stands for “Cloud of things”, is a research project run by industrial and research partners as well as city administrations from Eu- rope and Japan. The partners aim at developing infrastructure, services, tools and applications for municipalities and their stakeholders (citizens, ser- vice developers, etc.) to create, deploy, and manage user-centric applications based on loT and Cloud integration. Target applications include enhanced public transportation, increased citizen participation through mobile devices 30 Table 4: Platforms, services and research projects. $ E ali al; ES ES S|% 3 E É£ < ToTCloud Y [Y ]|na] n/a Y Y Y | Oct. 14 | Y OpenloT Y |vY Y Y Y Y Y Oct. 14 Y loT Toolkit [| Y | Y | n/a | n/a Y Y | n/a | Dec. 13 NimBits Y | Y Y partly Y Y n/a | Nov. 14 Y openPicus | Y n/a | un/a [n/a| n/a | Y n/a n/a “Iv Iv FA Y n/a FA Open.Sense | Y | Y | Y Y Y n/a Y ThingSpeak | Y | Y Y Y Y n/a Y CloudPlugs [| Y | Y | Y Y Y n/a Y Carriots viv | v Y Y n/a Y vv | Y Y Y | m/a | Oct. 14 | Y Y Y Y Y Nov. 14 Synapse loT Cloud Y Y Y Nov. 14 Y ClouT Y“ Iv]|v Y n/a | n/a | Y n/a (e.g. to photograph and record situations of interest to city administrators), safety management, city event monitoring, and emergency management. loT6 is a European research project on the future Internet of Things. It aims at exploiting IPv6 and related standards (e.g., GLOWPAN, CORE, COAP) to overcome current shortcomings (e.g. in terms of fragmentation) in the area of loT research and development. Its main objectives are to research, design, and develop a highly scalable IPv6-based service-oriented architecture to achieve interoperability, mobility, Cloud computing integra- > distribution among heterogeneous things, applications, tion, and intelligenc and services. The OpenloT project, cited before for its open source platform, aims at creating an open source middleware for getting information from het- erogeneous things, hiding the differences among these objects. The project explores efficient ways to use and manage Cloud environments for things and (such as sensors, actuators, and smart devices) and offering utility- pay-as-you-g0) loT services. Authors of [106] present a federation > Internet of Things loT-LAB (FIT loT-LAB) integrated with Ope- nloT, providing a very large scale infrastructure facility suitable for testing sensor devices and heterogeneous communicating objects. small wirel 31 Several projects target research issues related to loT and do not explicitly mention issues related to their integration with the Cloud. However, we report them here because they often mention data collection and elaboration platforms that are very likely being Clouds (right now or in the next years). These project include Smart Santander [112], The Cooperative ITS Corridor from Rotterdam to Vienna [30], and WISEBED [120]. 7. OPEN ISSUES AND FUTURE DIRECTIONS Thanks to the analyses we have done in Sec. 3, in Sec. 4, and in Sec. 5, here we resume the main issues related to CloudloT still requiring research efforts and point out some future directions (7, 82]. 7.1. Open Issues Standardization. The lack of standards is actually considered as a big issue towards CloudloT by a large number of researchers. Currently most things are connected to the Cloud through web-based interfaces, which are able to reduce the complexity for developing such applications [109]. However, they are not specifically designed for efficient machine-to-machine communications and introduce overhead in terms of network load, delay, and data processing. Moreover, interoperability is still an issue, because both the Cloud and the Things implement non-standard heterogeneous interfaces [35]. Even though the scientific community has provided multiple contributions to the deploy- ment and standardization of loT and Cloud paradigms, a clear necessity of standard protocols, architectures and APIs is being demanded in order to facilitate the interconnection among heterogeneous smart objects and the creation of enhanced services, which realize the CloudloT paradigm [115]. Power and Energy Efficiency. Recent CloudloT applications involve fre- quent data transmission from the things to the Cloud, which, in turn, may rely on smartphones as gateway [83]. Such process quickly drains battery ca- pacity on both the things and the gateway limiting the continuous operation to 24 hours or less. The literature shows that, in the field strictly related to the integration of Cloud and loT, obtaining energy efficiency in both data processing and transmission is an important open issue. However, signifi- cant research effort has already been spent for what concerns the Cloud and loT separately. For handling such issue, several directions have been pro- posed: more efficient data transmission and compression technologies [38]; 32 ously available. For instance, the Cloud can be configured to perform a set of important tasks and services for mobile multimedia users and networks, ranging from assessing the video quality level and load balancing to mul- timedia transcoding and redundancy/error correction schemes. This novel mobile multimedia era imposes new challenges for networks, contents, ter- minals, and humans, and must overcome problems associated, for instance, with high congestion, low scalability, fast battery consumption, and poor user experience. CloudloT involves M2M communications among many hetero- geneous devices with different protocols [16], which depend on the specific application scenario. M2M communication can be seen as an advanced form of sensor networks, where the ultimate goal is to provide comprehensive con- nections among all smart devices. However, the communication framework from sensor networks faces difficulties in satisfying the requirements of recent scenarios such as ITS, where each smart device can play more than one of the possible roles: sensor, decision maker, and action executor [86]. M2M largely benefited from the advances in wireless communication technologies, such as wearable and implantable biosensors, along with recent developments in the embedded computing, intelligent systems, and Cloud computing ar- eas [93]. In the literature, a few realizations of M2M communications have also been proposed, leveraging for example, Bluetooth (IEEE 802.15.1), Zig- bee (IEEE 802.15.4), or WiFi (IEEE 802.11b and p) technologies. However, there is still no consensus on the network architecture of a general scenario for M2M communications. Managing the things in a uniform fashion in a heterogeneous scenario, while providing required performance still represents an open issue [21, 15]. The majority of applications do not involve mobility: in stationary scenarios, loT often adopts IEEE 802.15.4/6LoWPAN solu- tions [103]. On the other hand, scenarios such as vehicular networks mostly adopt IEEE 802.11p. However, being WiFi and Bluetooth the most widely used radio technologies for wireless networks, their adoption for loT appli- cations is increasing: they represent a cheaper solution, most mobile devices already support them (e.g., smart phones), and both standards are becoming more and more low power. In some other cases, when power constraints are less critical, GPRS is still used for Internet comnectivity, but it results in a very costly solution (e.g., multiple SIM cards are necessary) [103]. Re- cently, serious attention was attracted by the standardization progress of LTE-Advanced, and the impacts of introducing M2M communications into LTE-Advanced are now under considerable study in 3GPP [86]. We be- lieve that research on network communications for CloudloT is still needed in order to provide effective and efficient solutions. SLA Enforcement. CloudloT users require things-generated data to be transferred and processed according to application dependent constraints, which can be strict in case of critical scenarios. Guaranteeing a certain QoS level about Cloud resources might not be always possible for a single provider, thus relying on multiple Cloud providers might be necessary to avoid SLA violations. However, dynamically selecting the best combination of Cloud providers still represents an open issue because of costs, time and heterogeneity of QoS management support [79]. Storage. Storage solutions have been frequently considered in this paper. For example, we have already considered them as a driver for the integration of Cloud and loT. However, the literature considers this as a still open issue as current solutions may not provide the necessary support for future appli- cations. For example, the storage of data transferred from the things to the Cloud involves some engineering issues still requiring research efforts. While data has to be properly timestamped to enable server-side reconstruction and processing, transfers require proper timing in order to avoid excessive bursti- ness of network and processing load [17]. One possible direction to address such issues involves the introduction of predictive storage and caching [66]. Scalability and Flexibility. CloudloT requires efficient mechanisms to match collected data and events to appropriate applications and services. Providing flexible subscription schemas and events management while guar- anteeing scalability with respect to things and users is still considered an open issue [79]. 7.2. Future Directions In order to enable the full potential of CloudloT, additional research effort is expected in several directions: e Properly identifying, naming, and addressing things will be necessary to support both the huge number of things and their mobility. While IPv6 could be the proper solution, its large-scale adoption is still an ongoing process and additional research is necessary to both speed up this slow process in specific scenarios (e.g. ac networks) and to cope with new mobility and scalability requirements. e Solutions for detecting environmental changes based on loT data will enable the delivery of enhanced context-based services, helping to pro- vide the best service depending on the situation. Such opportunity 36 will incentivate the research of more effective algorithms for delivering personalized contents and ads. e Large scale support for multi-networking (e.g., multihoming, multi- path, multicast), connection handover and roaming will be mandatory for improving network reliability and guaranteeing continuous connec- tivity, QoS, redundancy, and fault tolerance. In this context solutions based on Software Defined Networking are also envisaged. e Many applications of CloudloT would benefit from efficient and flexi- ble mechanisms for creating logically isolated network partitions over globally distributed network infrastructures, which could be another important driver for research in network virtualization and software- defined networking fields. e Converging towards a common open service platform environment for providing APIs to develop third-party CloudloT-based applications will enable new business opportunities and drive research efforts in the direction of defining standard protocols, libraries, languages, and methodologies for CloudloT. 8. CONCLUSION The integration of Cloud Computing and Internet of Things represents the next big leap ahead in the Future Internet. The new applications arising from this integration — we called CloudloT- open up new exciting directions for business and research. In this paper, we surveyed the literature in order to identify the comple- mentary aspects of Cloud and loT and the main drivers for integrating them into a unique environment. Since the adoption of the CloudloT paradigm enabled several new applications, we derived the main research challenges of interest for each of them. We further analyzed such challenges in order to identify current research directions. Finally, we surveyed available platforms and projects by comparing their main aspects and identified open issues and future research directions in this field. Thanks to the CloudloT paradigm everyday life and activities will be po- tentially improved for everyone: smart cities will enable more efficient public 37 [15] Bernaschi, M., Cacace, F., Pescape, A., Za, S., 2005. Analysis and experimentation over heterogeneous wireless networks. In: Tridentcom. TEEE. [16 Bhattasali, T., Chaki, R., Chaki, N., 2013. Secure and trusted cloud of things. In: India Conference (INDICON), 2013 Annual IEEE. IEEE, pp. 1-6. [17] Biswas, J., Maniyeri, J., Gopalakrishnan, K., Shue, L., Eugene, P. J., Palit, H. N., Siang, F. Y., Seng, L. L., Xiaorong, L., 2010. Processing of wearable sensor data on the cloud-a step towards scaling of continuous monitoring of health and well-being. In: Engineering in Medicine and Biology Society (EMBC), 2010 Annual International Conference of the TEEE. IEEE, pp. 3860-3863. [is Bitam, S., Mellouk, A., 2012. ITS-cloud: Cloud computing for Intel- ligent transportation system. In: Global Communications Conference (GLOBECOM), 2012 IEEE. IEEE, pp. 2054-2059. [19] Bo, Y., Wang, H., 2011. The application of cloud computing and the internet of things in agriculture and forestry. In: Service Sciences (1JCSS), 2011 International Joint Conference on. IEEE, pp. 168-172. [20 Bonomi, F., Milito, R., Zhu, J., Addepalli, S., 2012. Fog computing and its role in the internet of things. In: Proceedings of the First Edition of the MCC Workshop on Mobile Cloud Computing. MCC *12. ACM, New York, NY, USA, pp. 13-16. URL http: //doi.acm.org/10.1145/2342509.2342513 [21] Botta, A., Pescapé, A., Ventre, G., 2008. Quality of service statis- tics over heterogeneous networks: Analysis and applications. European Journal of Operational Research 191 (3), 1075-1088. (22 Carriots, 2014. https: //www.carriots.com. (23 Castro, M., Jara, A., Skarmeta, A., March 2013. Smart lighting solu- tions for smart cities. In: Advanced Information Networking and Ap- plications Workshops (WAINA), 2013 27th International Conference on. pp. 1374-1379. 40 (24] (27) (28] (29] (80) [81 [82] (83] Cerqueira, E., Lee, E., Weng, J.-T., Lim, J.-H., Joy, J., Gerla, M., 2014. Recent advances and challenges in human-centric multimedia mobile cloud computing. In: Computing, Networking and Communications (ICNC), 2014 International Conference on. IEEE, pp. 242-246. Chao, H.-C., 2011. Internet of things and cloud computing for future internet. In: Ubiquitous Intelligence and Computing. Lecture Notes in Computer Science. Chen, S.-Y., Lai, C.-F., Huang, Y.-M., Jeng, Y.-L., 2013. Intelligent home-appliance recognition over loT cloud network. In: Wireless Com- munications and Mobile Computing Conference (IWCMC), 2013 9th International. TEEE, pp. 639-643. Christophe, B., Boussard, M., Lu, M., Pastor, A., Toubiana, V., 2011. The web of things vision: Things as a service and interaction patterns. Bell Labs Technical Journal 16 (1), 55-61. Cloud Project, 2014. http: //clout-project.eu/. CloudPlugs, 2014. http: //cloudplugs.com/. Cooperative ITS corridor, 2014. http://www.bmvi.de/ SharedDocs/EN/Anlagen/VerkehrUndMobilitaet/Strasse/ cooperative-its-corridor.pdf?__blob=publicationFile. Copie, A., Fortis, T.-F., Munteanmu, V. I., 2013. Benchmarking Cloud Databases for the Requirements of the Internet of Things. In: 34th In- ternational Conference on Information Technology Interfaces, ITI 2013. pp. 77-82. Corsar, D., Edwards, P., Velaga, N., Nelson, J., Pan, J., 2011. Short paper: addressing the challenges of semantic citizen-sensing. In: 4th International Workshop on Semantic Sensor Networks, CEUR-WS. pp. 90-95. Cvijikj, I. P., Michahelles, F., 2011. The Toolkit Approach for End-user Participation in the Internet of Things. In: Architecting the Internet of Things. Springer, pp. 65-96. 41 [34] Dash, S. K., Mohapatra, S., Pattnaik, P. K., 2010. A Survey on Appli- cation of Wireless Sensor Network Using Cloud Computing. Interna- tional Journal of Computer science 4 Engineering Technologies 1 (4), 50-55. [35] Dinh, H. T., Lee, C., Niyato, D., Wang, P., 2013. A survey of mobile cloud computing: architecture, applications, and approaches. Wireless communications and mobile computing 13 (18), 1587-1611. [36] Distefano, S., Merlino, G., Puliafito, A., 2012. Enabling the cloud of things. In: Innovative Mobile and Internet Services in Ubiquitous Com- puting (IMIS), 2012 Sixth International Conference on. IEEE, pp. 858- 863. [37] Dobre, C., Xhafa, F., 2014. Intelligent services for big data science. Future Generation Computer Systems 37, 267-281. [38 Doukas, C., Maglogiamnis, I., 2011. Managing wearable sensor data through cloud computing. In: Cloud Computing Technology and Sci- ence (CloudCom), 2011 IEEE Third International Conference on. TEEE, pp. 440-445. [39] Doukas, C., Maglogiannis, 1., 2012. Bringing iot and cloud comput- ing towards pervasive healthcare. In: Innovative Mobile and Internet Services in Ubiquitous Computing (IMIS), 2012 Sixth International Conference on. IEEE, pp. 922-926. [40 Dukaric, R., Juric, M. B., 2013. Towards a unified taxonomy and ar- chitecture of cloud frameworks. Future Generation Computer Systems 29 (5), 1196-1210. 41 European Commission, 2013. Definition of a research and innovation policy leveraging Cloud Computing and loT combination. Tender spec- ifications, SMART 2013/0037. (42 Evangelos A, K., Nikolaos D, T., Anthony C, B., 2011. Integrating RFIDs and Smart Objects into a UnifiedInternet of Things Architec- ture. Advances in Internet of Things 2011. [43 Evans, D., 2012. The internet of everything: How more relevant and valuable connections will change the world. Cisco IBSG, 1-9. 42 [64] [65] [66 [67 loT Toolkit, 2014. http: //iot-tool1kit.com/. Irwin, D., Sharma, N., Shenoy, P., Zink, M., 2011. Towards a virtual- ized sensing environment. In: Testbeds and Research Infrastructures. Development of Networks and Communities. Springer, pp. 133-142. Islam, M. M., Hassan, M. M., Lee, G.-W., Huh, E.-N., 2012. A survey on virtualization of wireless sensor networks. Sensors 12 (2), 2175-2207. Jararweh, Y., Al-Ayyoub, M., Benkhelifa, E., Vouk, M., Rindos, A., et al., 2015. Sdiot: a software defined based internet of things frame- work. Journal of Ambient Intelligence and Humanized Computing 6 (4), 453-461. Jeffery, K., 2014. Keynote: CLOUDs: A large virtualisation of small things. In: The 2nd International Conference on Future Internet of Things and Cloud (FiCloud-2014). Ji, Y.-K., Kim, Y.-L, Park, S., 2014. Big data summarization using semantic feture for iot on cloud. Kamilaris et al., A., 2011. The smart home meets the web of things. International Journal of Ad Hoc and Ubiquitous Computing. Kantarci, B., Mouftah, H., 2014. Trustworthy sensing for public safety in cloud-centric internet of things. Kantarci, B., Mouftah, H. T., 2014. Mobility-aware trustworthy crowd- sourcing in cloud-centric internet of things. In: Computers and Com- munication (ISCC), 2014 IEEE Symposium on. IEEE, pp. 1-6. Kapadia, A., Myers, S., Wang, X., Fox, G., 2011. Toward securing sensor clouds. In: Collaboration Technologies and Systems (CTS), 2011 International Conference on. IEEE, pp. 280-289. Kitchenham, B., 2004. Procedures for performing systematic reviews. Keele, UK, Keele University 33, 2004. Kumar, L. D., Grace, $. S., Krishnan, A., Manikandan, V., Chinraj, R., Sumalatha, M., 2012. Data filtering in wireless sensor networks using neural networks for storage in cloud. In: Recent Trends In Information [76] [80] [81] [82] Technology (ICRTIT), 2012 International Conference on. IEEE, pp. 202-205. Kumar, N., March 2013. Smart and intelligent energy efficient pub- lic illumination system with ubiquitous communication for smart city. In: Smart Structures and Systems (ICSSS), 2013 IEEE International Conference on. pp. 152-157. Kuo, A. M.-H., 2011. Opportunities and challenges of cloud computing to improve health care services. Journal of medical Internet research 13 (3). Lazarescu, M., 2013. Design of a wsn platform for long-term environ- mental monitoring for iot applications. Emerging and Selected Topics in Circuits and Systems, IEEE Journal on 3 (1), 45-54. Le, X. H., Lee, S., Truc, P. T. H., Vinh, L. T., Khattak, A., Han, M., Hung, D. V., Hassan, M., Kim, M., Koo, K.-H., Lee, Y.-K., Huh, E.- N., Jan 2010. Secured wsn-integrated cloud computing for u-life care. In: Consumer Communications and Networking Conference (CCNOC), 2010 7th IEEE. pp. 12. Le, X. H., Lee, S., True, P. T. H., Khattak, A. M., Han, M., Hung, D. V., Hassan, M. M., Kim, M., Koo, K.-H., Lee, Y.-K., et al., 2010. Secured wsn-integrated cloud computing for u-life care. In: Proceed- ings of the 7th IEEE conference on Consumer communications and networking conference. IEEE Press, pp. 702-703. Le Tw'n, A., Quoc, H. N. M., Serrano, M., Hauswirth, M., Soldatos, J., Papaioannou, T., Aberer, K., 2012. Global sensor modeling and con- strained application methods enabling cloud-based open space smart services. In: Ubiquitous Intelligence € Computing and 9th Interna- tional Conference on Autonomic $ Trusted Computing (UIC/ATOC), 2012 9th International Conference on. IEEE, pp. 196-203. Lee, G. M., Crespi, N., 2010. Shaping future service environments with the cloud and internet of things: networking challenges and service evolution. In: Leveraging applications of formal methods, verification, and validation. Springer, pp. 399-410. 46 [83] [90] (91] Lee, K., Murray, D., Hughes, D., Joosen, W., Nov 2010. Extending sen- sor networks into the cloud using amazon web services. In: Networked Embedded Systems for Enterprise Applications (NESEA), 2010 IEEE International Conference on. pp. 1-7. Li, F., Vógler, M., ClaeBens, M., Dustdar, S., 2013. Efficient and scal- able loT service delivery on Cloud. In: Cloud Computing (CLOUD), 2013 IEEE Sixth International Conference on. IEEE, pp. 740-747. Li, W., Zhong, Y., Wang, X., Cao, Y., 2013. Resource virtualization and service selection in cloud logistics. Journal of Network and Com- puter Applications 36 (6), 1696-1704. Lien, S.-Y., Chen, K.-C., Lin, Y., 2011. Toward ubiquitous massive ac- cesses in 3gpp machine-to-machine communications. Communications Magazine, IEEE 49 (4), 66-74. Liu, C., Yang, C., Zhang, X., Chen, J., 2014. External integrity verifi- cation for outsourced big data in cloud and iot: A big picture. Future Generation Computer Systems. Lóhr, H., Sadeghi, A.-R., Winandy, M., 2010. Securing the e-health cloud. In: Proceedings of the 154 ACM International Health Informatics Symposium. ACM, pp. 220-229. Madsen, H., Albeamu, G., Burtschy, B., Popentiu-Vladicescu, F., July 2013. Reliability in the utility computing era: Towards reliable fog computing. In: Systems, Signals and Image Processing (IWSSIP), 2013 20th International Conference on. pp. 43-46. Marchetta, P., Natale, E., Salvi, A., Tirri, A., Tufo, M., De Pasquale, D., 2013. Trusted information and security in smart mobility scenarios: The case of s2-move project. In: Algorithms and Architectures for Parallel Processing. Springer, pp. 185-192. Martirano, L., Sept 2011. A smart lighting control to save energy. In: Intelligent Data Acquisition and Advanced Computing Systems (IDAACS), 2011 IEEE 6th International Conference on. Vol. 1. pp. 132-138. 47 [113] [117] [118 [119] [120 [121] [122 [123] Spillner, J., MúLler, J., Schill, A., Jun. 2013. Creating optimal cloud storage systems. Future Gener. Comput. Syst. 29 (4), 1062-1072. URL http: //dx.doi.org/10.1016/j.future.2012.06.004 Subashini, S., Kavitha, V., 2011. A survey on security issues in service delivery models of cloud computing. Journal of Network and Computer Applications 34 (1), 1-11. Suciu, G., Vulpe, A., Halunga, S., Fratu, O., Todoran, G., Suciu, V., 2013. Smart Cities Built on Resilient Cloud Computing and Secure Internet of Things. In: Control Systems and Computer Science (CSCS), 2013 19th International Conference on. TEEE, pp. 513-518. Synapse Internet of Things Cloud, 2014. https://www. synapse-wireless.com/snap-components/iot. Tan, K.-L., 2010. What's next?: Sensor+ cloud!? In: Proceedings of the Seventh International Workshop on Data Management for Sensor Networks. ACM, pp. 1-1. ThingSpeak, 2014. https: //thingspeak.com/. Wang, C., Bi, Z., Xu, L. D., May 2014. lot and cloud computing in au- tomation of assembly modeling systems. Industrial Informatics, IEEE Transactions on 10 (2), 1426-1434. WISEBED project, 2014. http: //www.wisebed. eu. Xiao, Y., Simoens, P., Pillai, P., Ha, K., Satyanarayanan, M., 2013. Lowering the barriers to large-scale mobile crowdsensing. In: Proceed- ings of the 14th Workshop on Mobile Computing Systems and Appli- cations. HotMobile '13. ACM, New York, NY, USA, pp. 9:1-9:6. URL http: //doi.acm.org/10.1145/2444776.2444789 Xively, 2014. https: //xively.com/. Xu, Y., Helal, S., Scmalz, M., 2011. Optimizing push/pull envelopes for energy-efficient cloud-sensor systems. In: Proceedings of the 14th ACM international conference on Modeling, analysis and simulation of wireless and mobile systems. ACM, pp. 17-26. [124] Yan, L., Zhang, Y., Yang, L. T., Ning, H., 2008. The Internet of Things: [195] [126] [127 [128 [129 [130 [131 [132 [133 ] ) ] from RFID to the next-generation pervasive networked systems. CRC Press. Yao, D., Yu, C., Jin, H., Zhou, J., 2013. Energy Efficient Task Schedul- ing in Mobile Cloud Computing. In: Network and Parallel Computing. Springer, pp. 344-355. Ye, X., Huang, J., 2011. A framework for Cloud-based Smart Home. In: Computer Science and Network Technology (ICCSNT), 2011 Inter- national Conference on. Vol. 2. IEEE, pp. 894-897. Ye, X., Huang, J., Dec 2011. A framework for cloud-based smart home. In: Computer Science and Network Technology (ICCSNT), 2011 Inter- national Conference on. Vol. 2. pp. 894-897. Yun, M., Yuxin, B., 2010. Research on the architecture and key tech- nology of Internet of Things (loT) applied on smart grid. In: Advances in Energy Engineering (ICAEE), 2010 International Conference on. TEEE, pp. 69-72. Zaslavsky, A., Perera, C., Georgakopoulos, D., 2013. Sensing as a ser- vice and big data. arXiv preprint arXiv:1301.0159. Zhang, Q., Cheng, L., Boutaba, R., 2010. Cloud computing: state-of- the-art and research challenges. Journal of internet services and appli- cations 1 (1), 7-18. Zhao, F., 2010. Sensors meet the cloud: Planetary-scale distributed sensing and decision making. In: Cognitive Informatics (ICCI), 2010 9th IEEE International Conference on. IEEE, pp. 998-998. Zhou, J., Leppanen, T., Harjula, E., Ylianttila, M., Ojala, T., Yu, C., Jin, H., 2013. Cloudthings: A common architecture for integrating the internet of things with cloud computing. In: CSCWD, 2013. IEEE. Zhu, J., Chan, D., Prabhu, M., Natarajan, P., Hu, H., Bonomi, F., March 2013. Improving web sites performance using edge servers in fog computing architecture. In: Service Oriented System Engineering (SOSE), 2013 IEEE 7th International Symposium on. pp. 320-323. [134] Zikopoulos, P., Eaton, C., et al., 2011. Understanding big data: An- alytics for enterprise class hadoop and streaming data. McGraw-Hill Osborne Media. [135] Zissis, D., Lekkas, D., Mar. 2012. Addressing cloud computing security issues. Future Gener. Comput. Syst. 28 (3), 583-592. Alessio Botta is a postdoc at the Department of Computer Engineering and Systems of the University of Napoli Fed- erico II (Italy). He graduated in Telecommunications Engi- neering (M.S.) and obtained the Ph.D. in Computer Engi- neering and Systems, both at University of Napoli Federico TI. His research interests are in the area of networking and, in particular, in the area of network performance measurement and improvement, with a specific focus on wireless and heterogeneous systems. Alessio Botta has coau- thored more than 40 international journal (IEEE Communications Magazine, TEEE Transactions on Parallel and Distributed Systems, Elsevier Computer Networks, etc.) and conference (IEEE Globecom, IEEE ICC, IEEE ISCC, etc.) publications. He has served and serves several technical program com- mittees of several international conferences (IEEE Globecom, IEEE ICC, etc.) and he acts as reviewer for different international conferences (IEEE Infocom, etc.) and journals (IEEE Transactions on Mobile Computing, IEEE Network, IEEE Transactions on Vehicular Technology, etc.) in the area of networking. In 2010 he was awarded with the best local paper award at IEEE ISCC 2010. Walter de Donato received the M.S. degree in computer engineering and the Ph.D. in computer engineering and sys- tems from the from the University of Napoli Federico II, Italy, where he currently works as a Post-Doc. During his PhD he visited the College of Computing at Georgia In- stitute of Technology of Atlanta, Georgia, USA, where he founded the BISmark project. Since July 2012 he holds a research and de- velopment manager position at Seven One Solution. He has co-authored over 20 international journal (Communications of the ACM, IEEE Network, Elsevier Computer Networks) and conference (SIGCOMM, USENIX ATC, PAM, Globecom, ...) publications, and he is co-author of a patent. His current research interests include methodologies, techniques, and distributed