Artificial intelligence-based load optimization in cognitive Internet of Things

The Internet of Things (IoT) comprises smart objects capable of sensing, processing, and transmitting application-specific data. These objects collect and transmit a huge amount of correlated and redundant data due to overlapped sensing regions, causing unnecessary exploitation of spectral bands and load balancing issues in the network. As a result, time-critical and delay-sensitive data experience a higher delay, lower throughput, and quality of service degradation. To circumvent these issues, in this paper, we propose a model that is energy efficient and is capable of maximizing the spectrum utilization with optimal Device-to-Gateway configuration. Initially, the network gateways perform spectrum sensing for available channels using an energy detection technique and forward them to a cognitive engine (CE). The CE assigns the best available channels in the licensed band to the network devices for communication. Each channel is divided into equal-length time slots for the timely delivery of critical data. In addition, the CE calculates the load on each gateway and uses particle swarm optimization algorithm for optimal load distribution among the network gateways. Our experimental results show that the proposed model is efficient for the resource-constrained IoT devices in terms of packet drop ratio, delay, and throughput of the network. Moreover, the proposed scheme also achieves optimal Device-to-Gateway configuration with efficient spectrum utilization in the licensed band.