ABSTRACT
The current prompt modernisation has created many issues that affect our daily life, such as congestion in the roadways, pollution in the environment, energy usage, and concerns about public safety. Investigation on “smart cities, smart homes, smart industries” has emerged in response to these difficulties, with the goal of using various Internet of Things technologies. Recently, the emphasis on research has switched to finding efficient ways to manage the massive number of data that metropolitan environments routinely create. This includes information on traffic flow, air pollution, healthcare, and other facets of city life via physical and participatory sensing sources. The data has been processed and analysed using computational intelligence techniques, which have yielded priceless insights that increase citizens' comprehension of their immediate environment. Both traditional and deep learning techniques are one of these methods that have attracted a lot of interest from the research community. It has more potential than conventional methods despite being a more recent paradigm in computer intelligence. The results of our study show that deep learning's complexity and its wide variety of applications in smart cities present a number of difficulties for this young subject. The effectiveness of deep learning has to be improved, new and emerging paradigms need to be explored, knowledge from many sources needs to be integrated, and privacy protection needs to be ensured, among other areas that we have identified as needing further attention. We want to expand the field's understanding of genuinely distributed intelligence for smart cities, smart homes, and smart industries by tackling these future directions.
