An Artificial Intelligence Approach to Enabled Smart Service Towards Futuristic Smart Cities

Introduction

Over the last decade or so, it has been observed that smart cities are metropolitan residences projected to offer a quality of life by offering the quality of service through essential and highly sophisticated cyber-physical systems (CPS). CPS is categorized as available pervasive services for improving lifestyle quality (Cicirelli, 2017), (Silva, 2018). Smart cities, and real-time smart services realization, can be achieved with new developments of cyber-physical systems which are more complex than the present form of IoT services. While incorporating concepts like cross-domain IoT applications and smart city, it can be viewed as an integration of a huge number of individual applications of the different domains to simplify and serve the complex services in a simpler and smarter way Intiza (2017), Soursos (2016). For example, when a person meets with an accident or a medical condition on road, then he has to give a call for an ambulance or known person for immediate help, which may not be feasible in critical cases, where a person is not in a situation to make a call to inform about his situation. In such cases, smart vehicle and smart city applications collaborate with each other to ensure that the proper medical facility should be made available, and information is sent to close relatives as well. This can be achieved only when different IoT applications of smart cities are seamlessly integrated to talk to one another and collaborate to achieve the diversified goal. This idea had motivated many researchers around the globe to work in this direction to come up with a robust design for a cross-domain IoT real-time applications system – a true all-inclusive smart IoT system in a real sense. For smart city big data analytics, cloud infrastructure is the best solution, where data is collected from different devices (intelligent vehicles, sensors, surveillance cameras, traffic lights, smart buildings, smart homes, & smart grids and meters) from different domains and from different layers. However, to achieve real-time services, a Fog/Edge layer has to be incorporated between the end devices and cloud layer to avoid the delay at the cloud layer Lilu (2019). In smart cities, a very large number of IoT devices obviously will generate enormous data, and thus both the storage and management of data in the fog layer need to be carefully considered. Ideally, these distributed environments required a huge amount of computation and storage capacity to handle and cloud servers are best suited for such scenarios, but today’s smart city applications can’t tolerate the delay of centralized cloud servers. In order to balance the two parameters like computation required and service delay, the Fog layer is introduced in between the cloud and access layer. Fog nodes are deployed to collect the data from devices that generate data like deployed sensors and moving autonomous vehicles on a real-time basis and process it on a real-time basis and send it back to vehicles for quick action. Periodically these collected data send to cloud servers for big data analytics. Network bandwidth and other network related parameters plays major role in sending collected data to upper layer and upper layer to cloud which are not considered in this work. To improve the network efficacy software defined networking can be an effective solution Hussain (2019), Hussain (2020) even in vehicular network SDN plays a major role improve the vehicular network efficacy (Renuka, 2021), whereas in this work, a standard network features considered. In this manuscript, both the pertinent concepts of context and context awareness are utilized to manage the collected data at the fog node level in an effective manner. Both context and context awareness has been excessively used by several researchers in Roy (2018), T. Gu (2005), Behera (2020), M. Baldauf(2006), and Renuka(2018). The concept learned from context generation and context processing is utilized as a context instance, and context instances are generated at fog nodes from raw data collected from either vehicles or roadside units. An application request can be processed in a Fog node if the required context instances and required resources like computation (CPU), Memory (Tcam), and Storage (disk) are available within. In a dynamic environment, the Fog node’s resource status also keeps changing, and every time a Fog node may not satisfy every application’s request.