Dynamic Mobile Crowdsourcing Selection for Electricity Load Forecasting.

The continuous growth of mobile devices in recent years has created a variety of opportunities for people to utilize the crowdsourcing technique to execute various intelligent computing and processing tasks, e.g., electricity load forecasting. However, there are few research in the field of real-time and accurate forecasting of electricity load in a dynamic environment, and this leads to an unsatisfactory result when applying the forecasting method to the real environment. In view of this challenge, in this paper, we propose a dynamic mobile crowdsourcing selection method for electricity load forecasting considering the dynamic arrivals of both crowdsourcing tasks and candidate workers to help the crowdsourcing platform find the ideal workers for the crowdsourcing tasks. Concretely, in our method, a system model is firstly established to quantify the ability of candidate workers in executing the crowdsourcing task when both workers and tasks arrive at or leave the platform dynamically; afterwards, a dynamic worker selection method is proposed based on the ability threshold of workers and the task priority. Finally, through a set of simulated experiments, we validate the feasibility of our proposal in terms of effectiveness and efficiency when making accurate electricity load forecasting through mobile crowdsourcing technique.