Development of double auction mechanisms and platform for perishable supply chain trading (PSCT)

Abstract

Perishable supply chain trading (PSCT) is the process of buying, selling, transferring, or exchanging products, services, and/or information across a perishable supply chain. Perishable products such as flowers, fish and vegetables are traded through Dutch auctions traditionally. However, the limited space of an auction center restricts the capacity of Dutch auctions. This thesis proposes double auction-based solution to match supply and demand in PSCT.

Three typical scenarios are investigated. The first scenario considers multi-unit call auctions with volume discount bids for PSCT. Two call auctions with volume discount bids are designed, i.e., incremental volume discount auction (IVDA) and total quantity discount auction (TQDA). Since the winner determination problems in these two auctions are NP-hard, two heuristic algorithms are proposed. Computational analysis is conducted to evaluate the performance of IVDA and TQDA. The experimental results demonstrate that the proposed call auctions with volume discount bids improve the market value compared to call auctions with ordinary bids.

In the second scenario, multi-unit multi-attribute call auctions are considered for PSCT. Firstly, a single-unit multi-attribute call auction (SMA) mechanism is designed for the multi/single-unit environment under which each buyer bids for multi-unit of one item and each supplier asks for a single unit of one item (i.e., single-output restriction). Then, a multi-unit multi-attribute call auction (MMA) mechanism is designed for the multi-unit exchange environment with a single item by relaxing the single-output restriction. The SMA and MMA mechanisms achieve incentive compatibility (IC), individual rationality (IR) and budget balance (BB). The allocative efficiencies of the proposed mechanisms are examined via simulation. The results show that the SMA and MMA mechanisms realize a desirable level of efficiency.

The last scenario considers multi-unit online double auctions for matching supply and demand between suppliers and buyers dynamically in PSCT. At first, a multi-unit online double auction (MODA) mechanism is designed, which consists of a price schedule and a matching rule. Two valid price schedules are suggested, i.e., fixed price and exponentially weighted moving average (EWMA) price. The MODA mechanism ensures IC, IR, BB and feasible. Then, two fairly truthful and fairly efficient mechanisms are proposed, namely, TV-MODA mechanism and TR-MODA mechanism. Through numerical experiments, the EWMA-price MODA and TV-MODA are found to be suitable for real-world applications.

In order to implement the proposed double auctions, a cloud-based auction platform (CAP) is developed. The CAP demonstrates several key contributions. Firstly, a set of facilities is provided to support the operations and decision making of the auction planner, auctioneer, suppliers and buyers. Secondly, cloud computing is adopted to realize “pay-as-you-go” and “pervasive-to-access” deployment. Thirdly, several issues in the design, development and implementation of the CAP are addressed to guarantee scalability, extensibility and usability of the CAP.