Abstract Wireless Sensor Network (WSN) is suffering from ailments like a small lifetime of participating nodes, packet loss during delivery, an end-to-end delay etc. Static allocation of sensor nodes and gateways are suitable for the predefined, pre-planned WSN. The more we comprise sensor nodes, the more we need gateways to accomplish the whole IoT (Internet of Things) based scenario. These communication imprecisions have led the trendy IoT network to be self-motivated in nature. This article proposed a gateway to gateway load balancing solution for smart talk static defined WS cluster network to overcome these issues. We have explored the unusual behavior of WSN gateways with suitable theoretical expressions and applied dynamic gateway selection formula to reduce end-to-end delay as well as to increase packet delivery rate. The experimental scenario has been developed using sensor devices, Arduino Uno, Raspberry Pi 3, Amazon web services, MQTT broker and private cloud platform in our Mobile Cloud computing laboratory. Theoretical analysis shows that the occurrence of delay due to increasing number of packets and message payload has reduced compared to the existing approaches. Experimental results confirm the correctness of the proposed model.