|
A Bilateral Bidding Mechanism for Cloud-Based On-Demand Transport Services
ZHOU Le-xin, XU Hai-ping, LI Ye
2020, 28 (3):
201-212.
doi: 10.16381/j.cnki.issn1003-207x.2018.1765
Cloud-based on-demand transport service marketplaces are usually based on the take-it-or-leave-it trading mechanism, where passengers and drivers can only accept or reject prices offered by the systems. Due to a lack of consideration for user requirements in such systems, system-generated prices do not necessarily reflect different situations of the traders, such as a passenger’s urgency degree, the actual running cost, and a driver’s expected profit.Since private information related to user requirements is incredibly valuable for determiningthe reasonable prices of trips, to design a trading mechanism for transport services utilizing such information become increasingly important. In this paper, a bilateral auction mechanism is produced for cloud-based on-demand transport service markets, where passengers and drivers are allowed to submit their bidsindependently.Our approach takes both passengers’ and drivers’ personal valuation of transport services into consideration. When a trip distance is less than a maximal distance determined by the market maker, an initial rate with a fixed fee is applied; otherwise, a passenger must pay by a unit priceno less than the price charged by a driver. Two rounds of bidding processes is considered: bidding for the initial rate and bidding for the unit price. In either round, trading price is determined p* that is used by all winnersof an auction. For example, in the second round, suppose we have n passengers with bidding prices bi, i=1, 2, …,n, and m drivers is sorted with bidding prices sj, j=1, 2, …, m. The bidding prices of passengers and drivers into b1≥ b2≥ … ≥ bn and s1≤ s2≤ … ≤ sm, respectively. Then it is found the positive integer k, called the efficient number of trades, that satisfies the requirements of bk≥sk and bk+1 < sk+1, where 1≤k≤min(m, n). The lower bound price is defined as lb=max(sk,bk+1) and the upper bound price is defined as ub=min(bk, sk+1). The trading price can be determined as p*=(lb+ub)/2. Since there must be an overlapping between the two ranges [sk,bk] and [bk+1, sk+1], it is guaranteed that p* exists. To support analysis of our approach, a success-oriented bidder as a bidder who tries his/her best to win an auction with non-negative profit, and aprofit-oriented bidder is defined as a bidder who tries his best to get more profit even at a high risk of losing the auction. Speculative bid of traders will inevitably reduce the trading successful rate of a passenger or a driver when the increase of traders’ profit is limited. It is proved that honesty is a dominant strategy for all passengers and drivers with success orientation, and an approximate dominant strategy for all passengers and drivers with profit orientation when the number of traders tends to infinity. It is further performed experiments where passengers and drivers demonstrate their speculative behaviors. For example, by generating random numbers of passengers and drivers in simulated auctions, it is shown that a profit-oriented passenger may try to increase his/her utility by placing a speculative bid, namely underbid, at the risk of losing the auction. This is consistent with our previous analysis that underbid is not a good strategy for success-oriented passengers. In addition, with multiple speculative passengers and speculative drivers in each auction, how the successful rate and trading price may be affected is studied. The experimental results show that the trading price has no obvious changes for various speculation degrees and percentages of speculative traders. In summary, our proposed bilateral auction mechanismnot only satisfies certain desirable properties, such as individual rationality, but also ensuresallvaluable profitable trades between passengers and drivers be supported. Furthermore, by analyzing speculativebehaviors of traders, it is shown that this trading mechanism encourages traders to provide honest bids, which helps to optimize the resource distribution of the platform.
References |
Related Articles |
Metrics
|