Elsevier

Decision Support Systems

Volume 124, September 2019, 113094
Decision Support Systems

A novel GSP auction mechanism for ranking Bitcoin transactions in blockchain mining

https://doi.org/10.1016/j.dss.2019.113094Get rights and content

Highlights

  • A novel GSP auction mechanism is proposed for Bitcoin transaction confirmation game.

  • Quality scores and virtual fees are introduced to rank transactions.

  • The proposed GSP mechanism is superior to the currently adopted GFP mechanism.

  • Quality scores and virtual fees can help users save fees.

  • Virtual fees allow all transactions to be processed more efficiently in a uniform pipeline.

Abstract

Bitcoin is gaining ground in recent years. In the Bitcoin system, miners provide computing power to confirm transactions and mine blocks in pursuit of transaction fees, while users compete by bidding transaction fees for faster confirmation. This process is in essence analogous to online ad auctions, where advertisers bid for more prominent ad slots. Therefore, inspired by the Generalized Second Price (GSP) mechanism dominantly used in online ad auctions, we propose to adopt the GSP auction model in the Bitcoin transaction confirmation game. Also, we use weighted fees as the new ranking basis, which can be calculated by user-submitted fees, transaction size, quality scores and virtual fees accumulated from the waiting time. We show that the formulated static GSP transaction fee auction with complete information has a unique Pure Strategy Nash Equilibrium. Then, we discuss the impacts of quality scores and virtual fees on users' equilibrium fee decisions and payoffs. Finally, computational experiments are designed to validate our theoretical models and analysis. Our research findings indicate that this novel GSP mechanism is superior to the currently adopted GFP mechanism, and can help users save fees. Besides, quality scores and virtual fees are also proven to be effective on reducing users' paid fees. Moreover, the design of virtual fees allows all transactions to be processed more efficiently in a uniform pipeline, and the interests of transactions with and without associated fees are taken into consideration.

Introduction

Blockchain technology has attracted intensive research interests and witnessed phenomenal development in recent years [18,22,12], thanks to its desirable features including peer-to-peer decentralization, trustlessness and tamper-resistance [5, 29]. The first and most successful blockchain system so far is widely known as Bitcoin, which creates a multi-billion dollar online economy and opens the new cryptocurrency era [15]. Bitcoin uses the Nakamoto consensus protocol to secure and update its underlying ledger of linked blocks. Typically, new blocks are created via miners repeatedly solving proof-of-work mining puzzles, that is, finding a random number that satisfies specific difficulty requirements using a brute force approach [4]. This process is called mining. The miner who finally wins the consensus competition has the right to confirm his/her packaged transactions and record them into the new block. The new block will then be appended to the main chain of previously agreed blocks, and the miner will get paid the block reward (i.e., 12.5 bitcoins currently) and transaction fees [9, 19].

In this process, transaction fees play the key role as the economic incentive to stimulate miners contributing their computing power so as to confirm transactions [27]. As a result, revenue-maximizing miners will preferentially confirm those transactions with higher fees, forcing Bitcoin users to increase their transaction fees for faster confirmation, or otherwise queue and wait. Since both the block generation rate and the size of Bitcoin blocks are restricted by the predetermined designs, the transaction confirmation rate is also constrained, which forces users to be faced with fierce competition and in turn high required fees if there is a transaction surge.

Therefore, there is a critical need in the individual-level for Bitcoin users to optimize their transaction fees, and also in the system-level for Bitcoin blockchain to improve its transaction ranking mechanism, with the aim of reducing the inflating transaction fees and enhancing the system efficiency. However, several features make this transaction ranking problem unique and challenging. First, transaction fees should be decided by users simultaneously with transaction submissions; and once determined, it is almost impossible for them to adjust any more according to current rules. Second, users are competing for perishable block space rather than storable objects. If no user submits transactions or miners choose to strategically mine an empty new block, the block space will be wasted. Third, unlike in the centralized markets, miners have the right to confirm any group of transactions they like to maximize their revenues, and apparently different miners have different mining costs. From the research perspective, it is better to model the transaction ranking problem with the game-theoretic analysis [8].

Currently, a large majority of Bitcoin transactions are processed in a pipeline with an order defined by the “rank-by-fee” mechanism, and the payment rule defined by the “pay-its-bid” mechanism, which is known as “Generalized First Price (GFP)” auction. That is, transactions with higher fees usually will be confirmed first and users need to pay their submitted fees. This simple rule will lead users to pay unnecessarily high fees to maintain a desirable ranking for their transactions. Generally, the higher is the required transaction fee, the longer a transaction might reside in the memory pool before being considered dormant [3]. In particular, when the memory pool encounters transaction congestions, users need to pay extra high fees for faster confirmation [16]. For example, in December 2017, the Bitcoin memory pool was stuffed with over 180,000 transactions, which leads to severe transaction congestion. Under that situation, users even need to wait for several days to get their transaction confirmed, and the required transaction fee per transaction surpassed 5 dollars 1. Transaction fees usually account for only a small percentage of the bitcoins transferred by transactions. However, it is possible that transaction fees might reach or even exceed the trading bitcoins, especially in micro-payment scenarios. For this consideration, the exorbitant transaction fees resulting from the GFP mechanism will render the system uneconomical for micro payments [11, 25].

Moreover, the GFP mechanism has been proven to be unstable in many scenarios due to the dynamic environment [7]. It may encourage inefficient investments in gaming the system. Under certain conditions, users will be engaging in cyclical fee adjustments, and equilibrium fees may follow a cyclical pattern with price-escalating phases interrupted by price-collapsing phases [30]. As such, it generates volatile prices that in turn causes allocative inefficiencies. In the Bitcoin system, each block is very likely to be created by different miners, and unstable fees cannot guarantee reliable incentives for them to keep mining as well as confirming transactions. As such, the GFP mechanism will lead to inefficiency of managing the transaction confirmation in this perspective.

Due to the aforementioned inherent demerits, the GFP mechanism fails to be a well-applicable auction mechanism for the Bitcoin transaction confirmation game. In this paper, we are motivated to deal with these problems resulting from the currently adopted GFP mechanism in the Bitcoin system. Existing online ad auction practices, especially the sponsored search auctions (SSA), have provided us with good reference models. In the SSA market, advertisers bid for prominent ad slots, which is essentially similar to the auction process of the Bitcoin transaction fee. The GFP mechanism was the original design for SSA since 1998, but has been replaced by the Generalized Second Price (GSP) mechanism in 2002, due to its inefficiency and high volatility shown in the market practice. The GSP mechanism was introduced by Google, and it earns over 90% of its revenues from the keyword auctions based on the GSP mechanism each year. Currently, almost all the online ad auctions, including SSA, real-time bidding and header bidding, etc., adopt the GSP mechanism, and it has created a considerably large global market of about 88 billion dollars in 20172. The GSP mechanism has been proven to be much more user-friendly and less susceptible to gaming, and thus is more stable and has higher allocative efficiency [2]. Inspired by the evolution of the SSA market from the GFP mechanism to the GSP mechanism, we propose to apply the GSP mechanism in the Bitcoin system. Actually, it can be tailored to the unique characteristics of the Bitcoin transaction fee auction. GSP insists that users offer a single fee for each transaction; consequently, even though users participate in the multi-objective game, their valuations can be properly represented by one-dimensional types [7]. Different from online ad auctions, one fee per transaction can be sufficiently expressive to fully convey Bitcoin users' preferences, because they need to independently determine the transaction fee for each transaction.

In the Bitcoin system, active users contribute a lot to maintain the high-quality sustainable development of the system. In practice, accompanied by the number of active users decreasing in 2018, the price of bitcoin dropped, the market size shrunk and miners' revenues decreased3. This shows that active users can influence the long-term interests of both the system and miners. Therefore, we should view high of active users. It is reasonable to assume that users having good experience in Bitcoin transactions will participate more frequently. In the SSA market, the user experience improvement is realized through the implementation of a weighted factor named “quality score” since 2008. The quality score is used in ranking ads, in order to ensure that the priority ads are shown in more prominent positions so as to encourage the quality advertising content [28]. Typically, ads with higher quality scores are accompanied with more impressions, better positions, and lower prices. Analogously, we also propose to use quality scores to reward those premium active users, so as to improve their probabilities to get higher ranks or alleviate the pressures of some micro-payment transactions to a certain extent via reducing paid fees.

In practice, transaction fees are not mandatory to Bitcoin transactions. In early stages of Bitcoin, 50KB in each block was reserved for high-priority transactions without fees, and their ranks are mainly determined by the waiting time4. That means the transaction confirmations are conducted in two separate pipelines, where the one is for transactions with fees, and the other is for priority transactions without fees. However, as the average transaction size grows, the reserved 50KB becomes insufficient to deal with the confirmation of those priority transactions. Besides, since Bitcoin Core v0.12 was launched, the priority rule was not performed by default any more5. Then, some users with priority transactions deviate to submit fees aiming to compete for better ranks in the fee pipeline. If we keep the ranking rule in the fee pipeline regulated uniformly by transaction fees, on the one hand, the priority rule will not work any more, which is not in favor of the priority transactions' benefits; on the other hand, transactions originally with fees are faced with more fierce competition, and those users with low fees will be ranked lower and thus experience longer delay. Therefore, we consider to deal with these two kinds of transactions through one uniform pipeline, and propose a new ranking basis incorporating both transaction fees and the waiting time. Inspired by the coupon designed to compensate customers' waiting time in the service-producing process [6], we use the “virtual fee” accumulated from the waiting time to compensate the over-long retention of transactions.

To summarize, our major contribution in this paper is to design a novel GSP auction mechanism for the Bitcoin transaction confirmation game, and also propose a new transaction ranking basis calculated by user-submitted fees, transaction size, quality scores and virtual fees. Also, we analyze the equilibrium of our proposed mechanism, and investigate the impacts of quality scores and virtual fees on users' fee decisions as well as allocative results of the limited block space.

The remainder of this paper is organized as follows. Section 2 briefly reviews the related literature. Section 3 establishes the static GSP auction model for the transaction confirmation game with complete information, and analyzes its pure strategy Nash equilibrium (PSNE). Also, we discuss the influences of quality scores and virtual fees. Section 4 conducts computational experiments to validate our theoretical models and analysis. Section 5 summarizes this paper and discusses the future work.

Section snippets

Literature review

Currently, the research efforts devoted to understanding the transaction confirmation game in the Bitcoin system are still quite limited. In view that the fixed transaction fee is equivalent to setting a maximum block size instead [8], the auction-based mechanism could be a better choice. However, Lavi et al. [13] argued that Bitcoin's current fee market based on the GFP auction does not extract revenue well for miners when blocks are not congested. Also, according to Houy [8], if transaction

The basic model

The GSP auction process of Bitcoin transactions can be briefly described by Fig. 1. First, new transactions pending for confirmation with associated fees are submitted by users, and they will enter the memory pool. Then, miners scan the relevant information of these unconfirmed transactions, and select a group of preferred transactions ranked in top positions as their mining basis. In this paper, we use weighted fees to determine transactions' ranks, which can be calculated by user-submitted

Computational experiments

In this section, we randomly choose the real transactions in block #567948 as the dataset of our experiments. This block was mined at 04:13 PM on March 20, 2019, and it is a full block with the size of 1,258,958 bytes. It collects total transaction fees of 0.22994263 bitcoins (BTC) from 2589 transactions6. Keeping the user-submitted fees and the existing allocation result under the GFP mechanism unchanged, the GSP mechanism can lead a saving of 5.12% on the

Conclusions and future work

Considering unique features of transaction fee auctions in the Bitcoin system, we propose a novel GSP auction mechanism to be tailored to its practical requirements of dealing with the problems caused by the currently-used GFP mechanism, which has been maturely employed in the practical online ad auctions.

In our proposed GSP auction, we introduce quality scores and virtual fees accumulated from the waiting time to formulate the weighted fees as the transaction ranking basis. The proposed GSP

Acknowledgments

We gratefully acknowledge the funding supports from the National Natural Science Foundation of China (#61533019, #71702182).

Juanjuan Li received her B.S. and M.S. degrees in Economics from Renmin University of China in 2008 and 2010, respectively, and she is a Ph.D. student with Beijing Institute of Technology since 2017. Currently, she is an Assistant Professor with the State Key Laboratory for Management and Control of Complex Systems, Institute of Automation, Chinese Academy of Sciences, Beijing, China. Her current research interests include blockchain, computational advertising and business intelligence.

References (1)

    Cited by (0)

    Juanjuan Li received her B.S. and M.S. degrees in Economics from Renmin University of China in 2008 and 2010, respectively, and she is a Ph.D. student with Beijing Institute of Technology since 2017. Currently, she is an Assistant Professor with the State Key Laboratory for Management and Control of Complex Systems, Institute of Automation, Chinese Academy of Sciences, Beijing, China. Her current research interests include blockchain, computational advertising and business intelligence.

    Yong Yuan received the B.S., M.S., and Ph.D. degrees from the Shandong University of Science and Technology, Qingdao, China, in 2001, 2004, and 2008, respectively, all in computer software and theory. He is currently an Associate Professor with the State Key Laboratory for Management and Control of Complex Systems, Institute of Automation, Chinese Academy of Sciences, Beijing, China. He is also with the Qingdao Academy of Intelligent Industries, Qingdao. His current research interests include social computing, blockchain, computational advertising and smart contracts.

    Fei-Yue Wang received the Ph.D. degree in computer and systems engineering from Rensselaer Polytechnic Institute, Troy, NY, USA, in 1990. He joined the University of Arizona, Tucson, AZ, USA, in 1990, and became a Professor and the Director of the Robotics and Automation Laboratory and the Program in Advanced Research for Complex Systems. In 1999, he founded the Intelligent Control and Systems Engineering Center, Institute of Automation, Chinese Academy of Sciences (CAS), Beijing, China, under the support of the Outstanding Overseas Chinese Talents Program from the State Planning Council and “100 Talent Program” from CAS. In 2002, he joined the Lab of Complex Systems and Intelligence Science, CAS, as the Director, where he was the Vice President for Research, Education, and Academic Exchanges with the Institute of Automation from 2006 to 2010. In 2011, he was named as the State Specially Appointed Expert and Director of the State Key Laboratory for Management and Control of Complex Systems, Beijing, China. His current research interests include methods and applications for parallel systems, blockchain, social computing, parallel intelligence, and knowledge automation.

    Dr. Wang has been the general or program chair of more than 30 IEEE, INFORMS, ACM, and ASME conferences. He was the President of the IEEE ITS Society during 2005–2007, the Chinese Association for Science and Technology, USA, in 2005, and the American Zhu Kezhen Education Foundation during 2007–2008. He was the Vice President of the ACM China Council during 2010–2011, and chair of IFAC TC on Economic and Social Systems from 2008 to 2011. Currently, he is the President-Elect of IEEE Council on RFID. Since 2008, he has been the Vice President and the Secretary General of the Chinese Association of Automation. He was the Founding Editor-in-Chief of the International Journal of Intelligent Control and Systems during 1995–2000 and the IEEE ITS MAGAZINE during 2006–2007. He was the EiC of the IEEE INTELLIGENT SYSTEMS during 2009-2012 and the IEEE TRANSACTIONS ON ITS during 2009–2016. He is currently the EiC of the IEEE TRANSACTIONS ON COMPUTATIONAL SOCIAL SYSTEMS, and the Founding EiC of the IEEE/CAA JOURNAL OF AUTOMATICA SINICA and the Chinese Journal of Command and Control. He was elected as a fellow of IEEE, INCOSE, IFAC, ASME, and AAAS. In 2007, he was a recipient of the National Prize in Natural Sciences of China and was awarded the Outstanding Scientist by ACM for his research contributions in intelligent control and social computing. He was a recipient of the IEEE INTELLIGENT TRANSPORTATION SYSTEMS (ITS) Outstanding Application and Research Awards in 2009, 2011, and 2015, and the IEEE SMC Norbert Wiener Award in 2014.

    View full text