Efficient Data Prediction, Reconstruction and Estimation in Indoor IoT Networks

Abstract

In an indoor Internet of Things (IoT) Network such as warehouse, factory and stadium, etc., a large number of IoT devices such as sensors, actuators, robots and drones, etc., continuously capture raw data over a regular interval of time. The captured raw data is then wirelessly transmitted to the gateways, which can be considered as the storing gadgets. These devices can gather, aggregate and analyze sensor data before transmitting to the Cloud. Continuous transmission of similar type of data by the IoT devices can lead to data redundancy, excessive channel utilization and bandwidth usage, etc. To overcome these problems, we develop a Compressive Sensing based Data Prediction (CS-DP) model which predicts data at the gateways by learning the data pattern received from IoT devices. Since there is a possibility of data loss while in transmission from the IoT devices to the gateways and at the gateways due to buffer overflow and other reasons, we propose a Compressive Sensing based Data Reconstruction (CS-DR) mechanism instead of retransmitting the data by the IoT devices. In addition to these, we also propose a Compressive Sensing based Data Estimation (CS-DE) technique which can bring down the volume of the data to be communicated to the cloud and improve the channel and space utilization of the system. All the above Compressive Sensing techniques are deployed at the gateways instead of the IoT devices or the cloud. This improves the system performance in terms of delay, communication and computation complexity and makes the system realizable in practice. Based on real data-set provided by an indoor test-bed, we conduct extensive MATLAB simulations to compare the performance of our proposed schemes with respect to the state of the art.