The Internet of Things (IoT) is a primary source of big data nowadays. However, transferring and processing IoT data directly in the cloud creates issues like increased access time, especially for time-sensitive applications like smart cities, and challenges in secure data transfer. Additionally, the large data volume strains cloud resources, reducing processing efficiency. In this paper, we reviewed the latest scientific papers that proposed a three-layer architecture at the computing level, for improving the efficiency the computing of data in real-time. Then we proposed a new three-layer architecture focusing on the most important levels in terms of reducing access time, and scalability, and improving efficiency with data security during its transfer between layers by dividing it depending on the level of complexity of data.