E-waste prediction and optimal route selection using adaptive deep Markov random field and block chain

Purpose An efficient e-waste management system is developed, aided by deep learning techniques. Here, a smart bin system using Internet of things (IoT) sensors is generated. The sensors detect the level of waste in the dustbin. The data collected by the IoT sensor is stored in the blockchain. Here, an adaptive deep Markov random field (ADMRF) method is implemented to determine the weight of the wastes. The performance of the ADMRF is boosted by optimizing its parameters with the help of the improved corona virus herd immunity optimization algorithm (ICVHIOA). Here, the main objective of the developed ADMRF-based waste weight prediction is to minimize the root mean square error (RMSE) and mean absolute error (MAE) rate at the time of testing. If the weight of the bins is more than 80%, then an alert message will be sent to the waste collector directly. Optimal route selection is carried out using the developed ICVHIOA for efficient collection of wastes from the smart bin. Here, the main objectives of the optimal route selection are to reduce the distance and time to minimize the operational cost and the environmental impacts. The collected waste is then considered for recycling. The performance of the implemented IoT and blockchain-based smart dustbin is evaluated by comparing it with other existing smart dustbins for e-waste management. Design/methodology/approach The developed e-waste management system is used to collect the waste and to avoid certain diseases caused by the dumped waste. Disposal and recycling of the e-waste is necessary to decrease pollution and to manufacture new products from the waste. Findings The RMSE of the implemented framework was 33.65% better than convolutional neural network (CNN), 27.12% increased than recurrent neural network (RNN), 22.27% advanced than Resnet and 9.99% superior to long short-term memory (LSTM). Originality/value The proposed E-waste management system has given an enhanced performance rate in weight prediction and also in optimal route selection when compared with other conventional methods.