POF: Probability-Driven Opportunistic Forwarding for Bike-Sharing System

Abstract

The opportunistic networks and the bike-sharing systems have been attracting much research attention in these years. In this paper, the cycling trips and the buffers of bike stations are utilized to relay the large volume of data. The Markov chain is exploited to formulate and estimate the delivery ratio with the mathematical lattice model. Then, the probability-driven opportunistic forwarding (POF) scheme, which calculates the delivery potentials from the possible delivery paths, is proposed. The relay evaluations, which consider the proper sequence of cycling trips, are made to support the data forwarding decisions. Moreover, the strategies, i.e., the discard notification and the backward elimination, are exploited to drop the redundant replications of data as many as possible. The extensive simulation results show that the proposed scheme outperforms several benchmark algorithms, e.g., prophet, spray & wait, and JDER, in terms of the delivery ratio, the delivery latency, and the expected hop count. INDEX TERMS Opportunistic network, bike-sharing system, cycling trips, delivery potential.

Optimization Model and Algorithm for Dockless Bike-Sharing Systems Considering Unusable Bikes in China

“…As travelers randomly arrive at a place nearby the hotspot to rent or return bikes, using historical operational data, the system can establish a probability density function g(D i,t ) of random demands(i.e., rents)D i,t and a probability density function k(R i,t ) of random returnsR i,t [10]. In fact, returned bikes are possibly to be unusable as stated in literature [7], [8]. This study assumes that with the aid of advanced machine learning technologies, the system can predict the probability of a returned bike being usable or unusable.…”

“…Constraint (6) requires that every vehicle is used only once, in which especially z 0,0,q = 1 means that vehicle q is not assigned and stays at the center without being used in practice. Constraint (7) makes sure that the current number of usable and unusable bikes in a vehicle q from node i to node j in the network cannot exceed the capacity of vehicle q. Constraint (8) defines the relationship between decision variable x i,t and intermediate variables s i,j , which implies that the number of replenishing bikes dropped off at hotspot i is the difference between the numbers of usable bikes in the same vehicle from previous hotspot j to hotspot i and from hotspot i to the next hotspot j. Constraint (9) defines the relationship between decision variable y i,t and intermediate variables w i,j , which implies that the number of unusable bikes picked up at hotspot i is the difference between the numbers of unusable bikes in the same vehicle from hotspot i to next hotspot j and from the previous hotspot j to hotspot i. Constraints (10), (11), and (12) are the definitions of decision variables and intermediate variables.…”

“…Thus, the maintenance operations, including recalling unusable bikes, replenishing new or usable ones and routing the maintenance vehicles, play a critical role of the sustainable development of dockless bike-sharing systems. However, the existing literature mainly focus on the rebalancing or called as reallocation of usable shared bikes [8], [9]. The maintenance operations of unusable bikes have not been received much attention from researchers.…”

“…In the last decade, the operations management of bikesharing systems has received much attention and interests from both academic and industrial fields. Existing literature mainly focuses on the rebalancing of shared-bikes among fixed stations to tackle the random arrival demands and returns [8], [9]. In order to predict the number of demands or returns at each station in a future period, Jia et al, [10] and Li and Zhao [11] respectively propose two prediction methods based on historical riding data.…”