Cloud is moving to the edge for smart cities

Turning Point for IoT

Smart cities are considered "smart" as cities are incorporating Information and Communications Technology (ICT) into their infrastructure. Cities heavily influenced by ICT have created networking, caching, and computing innovations. The basic components are sensors connected to city infrastructure that collect data to either store and analyze on the edge and a cloud to host the data for further analysis, decision making, and decision enhancement.

Cloud computing has been popular to enable access to a shared pool of computing resources. Typically, the distance between the cloud and end devices is very large and therefore cannot provide guarantees to low latency applications. Fog computing and edge computing have resulted to deploy computing resources closer to end devices. This results in a better quality of service for applications that require intensive computations and low latency.

Cloud Compute

Cloud computing has enabled IoT applications by providing on-demand storage, processing capabilities, scalability, elasticity, multi-tenancy, and resource pooling in a cost-effective manner. As smart cities are moving towards real-time data, the inherent limitations of high-latency and processing time inefficiencies due to the large volume of data are increasing the cost. Data needs to be analyzed closer to the source of the data. Data captured by cameras can incorporate multiple smart city applications. This eliminates the need for diverse sensors. Video-based data has high-bandwidth and is very expensive to transport the data over communications networks to a cloud or centralized data center for processing.

Edge Compute

With various smart city applications, the challenge of data deluge can be resolved by edge computing. Through edge computing, automated analytical computation is performed on data received from sensors and network switches rather than by sending the data to a centralized data repository. Once the data is analyzed on the edge, parameters can be set up as to which data needs to be sent to the cloud. This decreases latency and enhances the decision-making process for sensor data. Edge computing is further advanced with the incorporation of artificial intelligence (AI) and machine learning (ML); these perform data patterns and anomaly detection to save on time and cost by pre-processing and decision-making capabilities performed closer to the devices.

The convergence of edge and cloud

Smart cities are complex, enormous IoT deployments with both edge and cloud as underlying building blocks. As applications become more complex and services more diverse, it takes a village of various technology vendors to sense, collect and act on the data. Having a convergence of edge and cloud compute with interoperability and open framework architectures will be essential for a successful smart city deployment. The convergence of edge and compute provides modularity, heterogeneity, and flexible coupling to incorporate AI/ML while enabling incremental and scalable deployments for cost optimization.

An example of a converged edge and cloud compute is traffic lights. A traffic junction typically incorporates multiple applications such as lighting, signal operation, violation detection, congestion sensors, real-time surveillance, and weather monitoring systems that are simultaneously operating. Currently, these applications are connected to a fiber network. With edge and cloud convergence, data from these sensors can be captured, stored, analyzed, and controlled on the edge gateway. Analyzed data from the edge is then sent to the cloud. The analyzed data from various traffic junctions across the city are stored in the cloud where further analysis such as predictive analytics is performed. As a result, applications for first responders, travel times, commercial logistics, and air quality index monitoring applications can be delivered more efficiently.