Optimal Delivery Task Scheduling For Crowdsourced Delivery System

With the rapid development of Internet technology,e-commerce business is becoming increasing popular and important.Recent research es showed that in China,more and more peoples choose the online shopping,wherein business-to-customer(B2C)and customer-to-customer(C2C)are two of the most important online shopping schemes.Nowadays,many researchers have shown that the most challenging issue in online shopping is no longer the problem of how to attract consumers to consume online,but the problem of how to improve the logistics system and hence enhance the quality-of-experience(Qo E)of consumers.In order to reduce the cost of online shopping and speed up the shopping delivery efficiency,many international enterprises such as Wal-Mart and DHL have adopted the crowdsourced logistics system.Crowdsourcing is a new resource management scheme based on the concept of shared economy,and often relies on online systems to match the resource surplus and demand.Crowdsourced logistics,as a typical application example of crowdsourcing,aggregates the idle human resources to create cost-effective logistics systems.Crowdsourced logistics is st ill a new concept in China and adopted only by a few companies.However,with the fast developments of online business and online-to-offline(O2O)scheme,it has been viewed as one of the most important logistics technologies in the future.Our focus in this work is to study the optimal delivery scheduling problem in crowdsourced logistics systems,which can provide meaningful insights for the practical implementation of a real crowdsourced logistics system.One of the most important problems in crowdsourced logistics is the dispatch scheduling,that is,how to assign the couriers to deliver different packages to different receivers,given the set of couriers and packages.In this dissertation,we study the dispatch scheduling problem in two different scen arios: the "last mile" of traditional logistics and the local short-time logistics.Based on mathmetical modelling technique,we formulate the dispatch scheduling as a joint matching and routing problem,where the matching sub-problem focuses on the assignment of packages to couriers,and the routing sub-problem focuses on the route scheduling of each courier.We perform simulations to analyze how the performance(under the optimal scheduling solution)changes with the system perameters such as the courier number and their moving speed.Simulation results show that the overall social welfare and the total number of served packages increase concavely with the courier number and moving speed,while the average waiting time of customers decrease with the courier number and moving speed.Based on the above research results(for a static system with given couriers and packages),we further study the dispatch scheduling problem in the practical dynamic system,where both couriers and package requests arrive dynami cally and randomly.Existing solutions often adopt the independent scheduling strategy,that is,whenever a package request arrivals,it will be scheduled to a particular courier,without considering the potential package requests and new couriers in the n ear future.Thus,these existing solutions are often not optimal.In this paper,we propose a novel delayed scheduling strategy,where multiple package requests in one delayed scheduling window will be scheduled jointly.From simulation results,we find th e impact of the delayed scheduling window length on the system performance,which can provide guidance for the implementation of the delayed scheduling strategy in the practical dynamic system.