30-09-2016, 03:01 PM
1457079490-synopsis2.docx (Size: 105.88 KB / Downloads: 4)
Abstract
The taxi-sharing system we propose here, accepts taxi passengers real-time ride requests sent from smart phones and schedules proper taxis to pick up them via ridesharing, subject to monetary constraints, time and capacity. Such a system is of significant social and environmental benefit, e.g., saving energy consumption and satisfying people’s commute. For example, when the ratio of the number of ride requests to the number of taxis is 6, our proposed system serves three times as many taxi riders as that when no ridesharing is performed while saving 11 percent in total travel distance and 7 percent taxi fare per rider. Hence, we devise a mobile-cloud architecture based taxi-sharing system. Taxi riders use the taxi-sharing service provided by the system via a smart phone App. The Cloud first finds candidate taxis quickly for a taxi ride request using a taxi searching algorithm supported by a spatio-temporal index. To select a taxi that satisfies the request with minimum increase in travel distance, a scheduling process is then performed in the cloud. A ride request generator is developed in terms of the stochastic process modeling real ride requests learned from the data set. Tested on this platform with extensive experiments, the proposed system demonstrated its efficiency, effectiveness and scalability.
Introduction
Taxi is an important transportation mode between public and private transportations, delivering millions of passengers to different locations in urban areas. However, taxi demands are usually much higher than the number of taxis in peak hours of major cities, resulting in that many people spend a long time on roadsides before getting a taxi. Increasing the number of taxis seems an obvious solution. But it brings some negative effects, e.g., causing additional traffic on the road surface and more energy consumption, and decreasing taxi driver’s income (considering that demands of taxis would be lower than number of taxis during off-peak hours). The real-time taxi-sharing problem consists of a data model, constraints, and an objective function.
We place our problem in a practical setting by exploiting a real city road network and the enormous historical taxi trajectory data. Compared to existing carpooling systems, our proposed ridesharing model considers more practical constraints which include time windows, capacity, and monetary constraints for taxi trips. In addition, our work proposes efficient searching and scheduling algorithms that are capable of allocating the “right” taxi among tens of thousands of taxis for a query in milliseconds.
Existing System
Increasing the number of taxis seems an obvious solution. But it brings some negative effects, e.g., causing additional traffic on the road surface and more energy consumption, and decreasing taxi driver’s income (considering that demands of taxis would be lower than number of taxis during off-peak hours)
Disadvantages
• Unfortunately, real-time taxi-sharing has not been well explored, though ridesharing based on private cars, often known as carpooling or recurring ridesharing, was studied for years to deal with people’s routine commutes, e.g., from home to work .
• In contrast to existing ridesharing, real-time taxi-sharing is more challenging because both ride requests and positions of taxis are highly dynamic and difficult to predict.
• First, passengers are often lazy to plan a taxi trip in advance, and usually submit a ride request shortly before the departure.
• Second, a taxi constantly travels on roads, picking up and dropping off passengers. Its destination depends on that of passengers, while passengers could go anywhere in a city.
Proposed system
To address this issue in exisiting system, we propose a taxi-sharing system that accepts taxi passengers’ real-time ride requests sent from smartphones and schedules proper taxis to pick up them via taxi-sharing with time, capacity, and monetary constraints (the monetary constraints guarantee that passengers pay less and drivers earn more compared with no taxi-sharing is used).
Here, we report on a system based on the mobile cloud architecture, which enables real-time taxi-sharing in a practical setting. In the system, taxi drivers independently determine when to join and leave the service using an App installed on their smartphones. Passengers submit real-time ride requests using the same .Each ride request consists of the origin and destination of the trip, time windows constraining when the passengers want to be picked up and dropped off. On receiving a new request, the Cloud will first search for the taxi which minimizes the travel distance increased for the ride request and satisfies both the new request and the trips of existing passengers who are already assigned to the taxi, subject to time, capacity, and monetary constraints. Then the existing passengers assigned to the taxi will be inquired by the cloud whether they agree to pick up the new passenger given the possible decrease in fare and increase in travel time. Only with a unanimous agreement, the updated schedules will be then given to the corresponding taxi drivers and passengers.
Advantages
• Our system saves energy consumption and eases traffic congestion while enhancing the capacity of commuting by taxis.
• Meanwhile, it reduces the taxi fare of taxi riders and increases the profit of taxi drivers.
The architecture of our system is presented in Fig. 1. The cloud consists of multiple servers for different purposes and a monitor for administers to oversee the running of the system. Taxi drivers and riders use the same smart phone App to interact with the system, but are provided with different user interfaces by choosing different roles. As shown by the red broken arrow (a), a taxi automatically reports its location to the cloud via the mobile App when (i) the taxi establishes the connection with the system, or (ii) a rider gets on and off a taxi, or (iii) at a frequency (e.g., every 20 seconds) while a taxi is connected to the system.We partition a city into disjoint cells and maintain a dynamic spatio-temporal index between taxis and cells in the indexing server depicted as the broken arrow (b).
Denoted by the solid blue arrow 1 , a rider submits a new ride request Q to the Communication Server. Represented by the blue arrow 4 , the communication server sends ride request Q and the received candidate taxi set SV to the Scheduling Server Cluster. The scheduling cluster checks whether each taxi in SV can satisfy Q in parallel (detailed in Section 5), and returns the qualified taxi V that results in minimum increase in travel distance and a detailed schedule, shown as arrow 5. Each rider R who has been already assigned to the taxi V will be enquired whether they would like to accept the join of the new rider, as depicted by blue arrow 6.
Conclusion And Future Enhancement
We proposed and developed a mobile-cloud based real-time taxi-sharing system. We presented detail interactions between end users (i.e. taxi riders and drivers) and the Cloud. The experimental results demonstrated the effectiveness and efficiency of our system in serving real-time ride requests.
Firstly, our system can enhance the delivery capability of taxis in a city so as to satisfy the commute of more people. Secondly, the system saves the total travel distance of taxis when delivering passengers. Thirdly, the system can also save the taxi fare for each individual rider while the profit of taxi drivers does not decrease compared with the case where no taxi-sharing is conducted. Using the proposed monetary constraints, the system guarantees that any rider that participates in taxi-sharing saves 7 percent fare on average.
In the future, we consider incorporating the creditability of taxi drivers and riders into the taxi searching and scheduling algorithms. Additionally, we will further reduce the travel distance of taxis via ridesharing. We also consider refining the ridesharing model by introducing social constraints, such as gender preference, habits preference