Coud computing has emerged during the last years as a disruptive model, with the ability to transform IT organisations, helping them to become more responsive and agile than ever before. Due to its multiple and significant benefits, cloud computing can particularly help complex organisations such as cities run more efficiently, providing new opportunities and opening up new business models. Although clearly at an early stage, the discussion on how cloud computing can help cities towards their effort of becoming smart has revealed a plethora of applications regardless of the size and level of organisation, and wealth of a city. The paper aims to contribute to this discourse both at the theoretical and empirical level, firstly, by reviewing the literature on the cloud computing paradigm and the way smart cities can benefit from it, secondly, by introducing a cloud- based solution for smart city services and thirdly, by doing a short foresight on future research challenges with relation to cloud computing adoption by smart cities. The cloud computing model represents a fundamental change in the way that information technology hardware and software are created, developed, deployed, scaled, updated, maintained and paid . It serves as an enormous step towards delivering global computing as a utility (like traditional utilities such as water, electricity and telephony) by changing the traditional access model, where data and applications are fully contained in the same physical location (local computing), to a new one, where the users access their data and applications outside their own local computing environment through the internet.
Cloud computing promises economic benefits, speed, agility, flexibility, rapid elasticity and more innovation. The motivations of the organisations for the migration of their applications to the Cloud are closely related to the following key cloud computing characteristics:

A consumer can unilaterally provision IT resources (e.g., storage, processing power, memory, bandwidth, etc.), as needed automatically without requiring human interaction with the cloud provider.

IT resources are available over the Internet and accessed through heterogeneous devices (e.g., mobile phones, tablets, laptops, and workstations). There is a sense of location-independence because the customer generally is not aware of the exact location of the provided resources.

A cloud environment offers to the consumer the ability to rapidly scale up or down the IT infrastructure commensurate with demand. To the consumer, the capabilities available for provisioning usually appear to be limitless and can be reserved in any quantity at any time.

Cloud systems monitor, control and report the use of IT resources by leveraging a metering capability at some level of abstraction suitable for the type of service. This leads to a transparent relationship between the consumer and provider of the cloud service. The above-mentioned characteristics create a highly efficient, scalable and elastic computing environment, which is available through a business model where consumers buy only the capacity and capabilities needed at any given time, instead of buying and deploying all the components of a computing centre. Using this “pay for what you consume” approach, an organisation can significantly reduce the up-front costs by avoiding the procurement of hardware and software in advance, as well as the resultant infrastructure depreciation. Moreover, cloud adoption can decrease the operational costs, as the organization will maintain at a given time only the required resources, which could be scaled up and down through rapid elasticity. Also, other IT costs can be reduced by buying general purpose capabilities such as asset management, security, collaboration, etc. as a service, instead of maintaining a specialised in-house IT department. As organisations operate in an unstable economic environment, smart consumption-based procurement allows them to scale up to fulfil new demands and reduce spending, if necessary, to address changes in budgets and funding.

City governments and municipalities everywhere constitute for one thing complex public organisations that have more reasons to invest in cloud computing than any other public organisation. Despite their limited resources they have to provide a wide number of municipal services (ranging from sanitation, water, schools, health, transportation etc.) and serve the needs of their citizens in their daily life. At the same time, they face a variety of challenges including job creation, economic growth and environmental pollution. While it is widely accepted that increasing urbanisation strains the limited resources of cities and affects their resilience, it also highlights the need for sustainable urban development; especially in terms of much more efficient management of natural resources, such as energy and water, as well as better planning and collaborative decision making . In this context, cloud computing can play a significant role, facilitating cities in meeting the above-mentioned tasks. Over the past years, the term ‘smart cities’ has evolved to denote the cognitive processes combined with the deployment of ICTs, institutional settings for innovation and physical infrastructure, which taken altogether increased the problem solving capability of a city or a community . The main features of a smart city include applications that connect, manage and optimise data from a complex set of devices, sensors, people and software, creating real-time, context-specific information intelligence and analytics, which aim to transform the urban environment and address its specific needs . Managing such enormous amount of heterogeneous data requires, among others, high storage capacity and performance computing power. For this, the latest developments in cloud computing and the Internet of Things (IoT) are widely deployed in smart cities. More specifically, smart cities have to use a wide variety of ICT solutions to deal with urban problems and monitor their functions; they do not only require the use of new technologies and devices such as sensors, RFID (radio-frequency identification) devices, smartphones, smart household appliances, etc., to collect land use, transport, census, and environmental monitoring data which are generated every minute in the urban environment, but also the capacity to manage and process all this large scale data (Big data) in real time, in an interconnected and service/applications’ specific way . The emergence of cloud computing paradigm facilitates big data storage and big data integration, visualisation, processing and analysis in acceptable time frames. Fu, Jia and Hao listed six reasons for the convergence of cloud and IoT in smart cities:

• the immense storage capacities of cloud computing infrastructure,
• central processing capacity to perform complex computing,
• the ability for dynamic reconfiguration of resources, securing sufficient computational resources at any time,
• the ability to conduct system level comprehensive analysis with sample data,
• high accessibility to various objects in IoT through user friendly applications and customised portals and
• high speed network and disaster recovery capabilities. Most significantly, cloud based big data mining and analytic tools can deal effectively with multidisciplinary city data, characterised by environmental dynamism and spatiotemporal attributes, and formulate a variety of smart city application scenarios

Based on the advances made so far, this section aims to highlight new scientific directions and future challenges with regards to smart cities and cloud computing. Although the trends that define the future of cloud computing can be numerous, ranging from technological aspects to new business models/opportunities, we identified four areas that are about to play a significant role with regards to cloud computing adoption from municipalities and city governments. As Petrolo, Loscri and Mitton mentioned, “in Smart City context, Cloud of Things (CoT) is expected to play a significant role in making a better use of distributed resources, achieving higher throughput and tackling large scale of computational problems, to enable the horizontal integration of various vertical IoT platforms and the Smart City vision”. This means that over the next years, the focus will be on cloud platforms dedicated for IoT and on technologies for real time processing of big data and linked data. Advanced analytics over millions of data streams coming from highly distributed, heterogeneous, decentralised, real and virtual devices and data sources which bring a new vision on the notion of cloud scalability. Here, issues of interoperability, privacy and security should be carefully considered. The SCP could be expanded in order to enable the rapid deployment of city wide networked sensors and actuators, as well as of IoT applications. Three new modules should be developed to support the core requirement of the IoT solutions: • The Control Module that will handle the device interfacing and will enable the time-critical response. The module will consist of autonomous control applications and drivers which support a broad range of existing and new equipment and protocols. • The Data Provisioning Module that will provide a unified sensor data acquisition in order to enable further processing steps. The module will store the acquired data in the SCP Data Service Layer. Data transformation techniques, such as data masking and data obfuscation, will be implemented in order to address privacy and security issues. • The Big Data Analytics Module that will provide historic as well as real-time analytics. The historic analytics will identify data patterns of significance and will enable the optimisation of algorithms, services, and solution delivery while the real-time analytics will evaluate data as they come into the system in order to produce insights in near-real- time for immediate exploitation. Apart from the development of the new modules, the platform’s Data Service Layer should be enhanced with big data handling capabilities.