
SoK: Diving into DAG-based Blockchain Systems Qin Wangy,x, Jiangshan Yuz, Shiping Chenx, Yang Xiangy y Swinburne University of Technology, Melbourne, Australia. z Monash University, Melbourne, Australia x CSIRO, Data61, Sydney, Australia ABSTRACT scalability, cannot perfectly co-exist in a blockchain system. For Blockchain plays an important role in cryptocurrency markets example, one major hindrance to Nakamoto Consensus (NC) [1] and technology services. However, limitations on high latency and is the reduced security with an increased block generation rate. low scalability retard their adoptions and applications in classic The security of NC’s longest chain wins rule requires that honest designs. Reconstructed blockchain systems have been proposed nodes should be aware of other’s blocks soon after the block’s pro- to avoid the consumption of competitive transactions caused by duction. If the block creation rate exceeds the propagation time, linear sequenced blocks. These systems, instead, structure trans- concurrent blocks increase the possibility of forks happening. To actions/blocks in the form of Directed Acyclic Graph (DAG) and mitigate the bottleneck of performance, multiple approaches have consequently re-build upper layer components including consensus, been proposed from different angles. incentives, etc. The promise of DAG-based blockchain systems is to Potential Solutions. These approaches include the methods on enable fast confirmation (complete transactions within million sec- sharding technique [11][12], layer2 protocols [13], sidechain tech- onds) and high scalability (attach transactions in parallel) without nique [14], heterogeneous structure [15], hybrid consensus solu- significantly compromising security. However, this field still lacks tions [16] and assisted techniques [17] such as modifying hard- systematic work that summarises the DAG technique. To bridge coded parameters [18] and cross-chain technique [19]. Sharding the gap, this Systematization of Knowledge (SoK) provides a com- splits pending transactions into smaller shards and makes them prehensive analysis of DAG-based blockchain systems. Through processed in parallel, but it is hard to achieve the consensus across deconstructing open-sourced systems and reviewing academic re- shards due to the asynchronization. Cross-chain protocols help to searches, we conclude the main components and featured proper- partially solve the problem by establishing the channels between ties of systems, and provide the approach to establish a DAG. With multiple shards (horizontal sharding [19]). As a sacrifice, these on- this in hand, we analyze the security and performance of several top protocols inevitably result in heavy and redundant systems, leading systems, followed by discussions and comparisons with con- weakening efficiency and stability. Layer2 protocols enable par- current (scaling blockchain) techniques. We further identify open ticipants to perform off-(main)chain transactions through private challenges to highlight the potentiality of DAG-based solutions and communication rather than broadcasting to the whole network. indicate their promising directions for future research. Together with such features, the challenge is how to properly and effectively guarantee the validity and consistency between off-chain KEYWORDS and on-chain transactions. The technique of sidechain pegs the aux- DAG-based Blockchain, SoK, Performance iliary chains to involve more transactions. Heterogeneous structure adds the new types of blocks to assign them with different tasks. 1 INTRODUCTION Hybrid consensus solution combines multiple fundamental consen- Limitations. Blockchain becomes an emerging technology to real- sus mechanisms together (PoX+BFT [15][16][20]) to integrate their ize the distributed ledgers1. The rising interest in blockchain has benefits. Modifying hard-coded parameters e.g. directly increases attracted extensive attention. Thanking to its great potential to the volume of block from 1M to 8M in BCH [18]. However, these tackle critical security and trust challenges in various distributed techniques are still based on the linear-based backbone protocol, environments, blockchain technology enjoys rapid development limiting the scope of exploitation. Therefore, the radical modifi- arXiv:2012.06128v2 [cs.CR] 14 Dec 2020 [1][2] with derived topics evolving into a well-studied field [3][4] cation – reconstructing chains from the underlying structure and in both industry [5][6] and academia [7][8]. However, the great topology, become an emerging solution. impact of blockchain promotes an influx of participants joining the DAG-based Approach. Linearly structured blockchain systems game. Continuously increased traffic results in unavoidably cata- maintain all the transactions/blocks in one single chain. Concurrent strophic congestion due to the performance bottlenecks including transactions/blocks compete for one valid position each round. This slow confirmation, low throughput and poor scalability. Compared to design inevitably leads to slow confirmation due to competitive min- most centralized systems, these factors cannot be easily improved ers, conflicted transactions, and wasted computations. The perfor- in blockchain systems which require decentralization as the prior- mance of the system can only be artificially suppressede.g. ( adjust ity. Several studies raise the view of blockchain trilemma [9] and the confirm time) so that each block is fully attached before thenext the trade-off [10], claiming that the decentralization, security, and one’s arrival. Very few orphan blocks can be involved in the system. 1In this paper, we regard “blockchain” as a general term covering types and models To this end, aiming to enable more transactions simultaneously pro- which are based on the same technology. For simplicity, we ignore the difference cessed/confirmed motivates the emerge of DAG-based blockchain between “blockchain” and “distributed ledger”. Also, we occasionally use “DAGs" to represent “DAG-based Blockchains” for short. systems [21][22][23][24]. They structure the transactions/blocks in the form of graph topology to underlyingly alter the actual oper- sections enable a deep understanding of the design princi- ations. DAG-based systems can improve performance by requiring ples, system features, and unique properties. Readers can less communication, computation and storage overhead. However, learn the full views of collected systems through these multi- the sources of existed open-sourced projects and solutions are dis- dimensional analysis. parate and disorganized. Various basic concepts in DAGs (e.g. vertex ⋄ Discussion. Additionally, we review concurrent techniques meaning, consensus approach, order sequence) are still confused, surrounding the DAG and clarify a series of open questions and potential challenges have neither been clearly identified. An in Section 8. Detailed discussions help us learn more about organized and structured systematic overview is absent for newcom- current challenges and the potential future of this field. ers. Several studies draw their attention to this field, but the works ⋄ Summary. As a summary, our analysis framework has an- are either rough in summaries [25][26][27][28][29][30][31][32], swered the listed questions in detail. Discussions on elements superficial in analysis [17][33][34][35][36][37][38][39][40], or in- of the DAG model provide us a simple but complete classi- complete in evaluations [41][42][43][44]. fication of current DAG-based systems, answering Q1 and Q2. Deconstructions of consensus mechanisms show us how Contributions. In this SoK, we attempt to consolidate the core current DAG-based systems operate with their unique de- knowledge of the structural shift in blockchain systems and review signs, answering Q3. Analysis of the properties, securities, the state-of-the-art DAG-based blockchain systems with compre- and performance enables us to understand Q4, Q5, and Q6. hensive mechanisms and properties. To drive future research, we Discussions, comparisons, and challenges answer Q7 and also provide multifaceted discussions and comparisons with concur- Q8. Surrounding details about the origination, progress, and rent scaling techniques of blockchain. Furthermore, we summarise application are provided in Section 8, 9. unsolved problems existed in current DAG systems, hoping to high- light the research challenges in this field. We start with a series of Insights. Our work provides a roadmap for studying the applied- simple questions, aiming to provide brief guidance for readers. DAG systems. Based on our scrutinized analysis, we provide several insights in this part: a) DAG-based blockchains are still far away Q1. What is the DAG-based blockchain. from commercial applications due to their incompatible designs, Q2. How do they structure the ledgers. absence of standards, unreliable security, varied performance, and Q3. How to run the consensus. unfinished implementations. b) DAG structure has the potential Q4. What are the desired properties. to improve scalability and performance but inevitably sacrifices Q5. How secure are they. certain properties like consistency or finality. c) DAG-based sys- Q6. What is the performance improvement. tems vary from one to another. A uniformly formalized model can Q7. How is this different from other techniques. hardly cover all key points. Instead, a loose and informal model Q8. What are the research challenges. benefits a lot for better understanding. d) Although applying
Details
-
File Typepdf
-
Upload Time-
-
Content LanguagesEnglish
-
Upload UserAnonymous/Not logged-in
-
File Pages36 Page
-
File Size-