Source-linked AI summary
A Survey on the Security of Blockchain Systems
Xiaoqi Li, Peng Jiang, Ting Chen, Xiapu Luo, Qiaoyan Wen
TL;DR
Blockchain security lacks a systematic examination spanning risks, real attacks, and security enhancements across popular systems. This paper studies those dimensions, finding concrete vulnerabilities and attacks while synthesizing practical protections and future directions.
Problem
Prior studies did not systematically examine blockchain-system risks, corresponding real attacks, and security enhancements across popular blockchain systems.
Method
The paper studies popular blockchain systems, analyzes security risks and exploited vulnerabilities in real attacks, and reviews security enhancements and future directions.
Results
The survey documents concrete blockchain vulnerabilities and attacks, including privacy weaknesses, smart-contract gas-costly patterns, and major financial losses.
Takeaways & Limitations
The reviewed security enhancements and suggested future directions provide material for developing more secure blockchain systems.
Abstract
from arXiv · showhide
Since its inception, the blockchain technology has shown promising application prospects. From the initial cryptocurrency to the current smart contract, blockchain has been applied to many fields. Although there are some studies on the security and privacy issues of blockchain, there lacks a systematic examination on the security of blockchain systems. In this paper, we conduct a systematic study on the security threats to blockchain and survey the corresponding real attacks by examining popular blockchain systems. We also review the security enhancement solutions for blockchain, which could be used in the development of various blockchain systems, and suggest some future directions to stir research efforts into this area.
1. Introduction
Blockchain has broad adoption but significant security incidents and vulnerabilities motivate a systematic study. The paper examines risks, real attacks, security enhancements, and future directions across popular blockchain systems.
- Blockchain has expanded from Bitcoin into applications including medicine, economics, Internet of Things, and software engineering.
- Security vulnerabilities can cause major financial losses, including approximately 60 million dollars stolen from the DAO and 450 million dollars in Bitcoin stolen from MtGox.
- Existing studies did not systematically examine blockchain-system risks, corresponding real attacks, and security enhancements across popular blockchain systems.
- The paper presents a systematic examination of security risks to popular blockchain systems.
- It surveys real attacks from 2009 through May 2017, analyzes exploited vulnerabilities, summarizes practical security enhancements, and suggests future research directions.
2. Overview of Blockchain Technologies
Blockchain technologies use decentralized consensus and block propagation to maintain shared data, while evolving from cryptocurrency systems toward smart-contract platforms. The overview covers consensus mechanisms, synchronization, and the development of blockchain applications.
- Consensus Mechanism: Blockchain uses decentralized consensus mechanisms to support data and transaction reliability without a third-party trusted authority.
- Consensus Mechanism: PoW requires nodes to solve computationally hard but easily verifiable puzzles before broadcasting blocks for consensus.
- Consensus Mechanism: In Bitcoin, a block includes PrevHash, nonce, and transactions; the nonce must satisfy SHA256(PrevHash || Tx1 || Tx2 || ... || nonce) < Target.
- Consensus Mechanism: PoS replaces PoW computation with cryptocurrency ownership, returning deposits as bonuses for valid blocks or transactions and fining invalid ones.
- Block Propagation and Synchronization: Propagation mechanisms progress from advertisement-based exchange toward sendheaders, unsolicited push, relay-network, and Ethereum’s push/advertisement hybrid approaches.
- Block Propagation and Synchronization: Block synchronization locates a common ancestor through header requests and then requests blocks from that ancestor, with actual transfers potentially below MaxHeaderFetch.
- Technology Development: Blockchain development progressed from cryptocurrency-focused blockchain 1.0 to blockchain 2.0, where smart contracts enable decentralized applications.
3. Risks to Blockchain
The paper organizes blockchain risks into nine categories and distinguishes risks shared by blockchain 1.0 and 2.0 from risks specific to smart-contract systems.
- The paper divides common blockchain risks into nine categories and examines their causes and possible consequences.
- Risks in blockchain 1.0 and 2.0 are mostly related to blockchain operation mechanisms.
- Risks introduced by blockchain 2.0 are unique to smart-contract systems and usually arise from contract development, deployment, and execution.
3.1. Common Risks to Blockchain 1.0 and 2.0
Blockchain 1.0 and 2.0 face shared risks involving consensus, private-key security, illicit use, double spending, and transaction privacy. These risks include both protocol-level attacks and weaknesses in privacy-preserving mechanisms.
- Consensus and account security: A miner controlling more than 50% of PoW hashing power can reverse transactions, double-spend, reorder transactions, disrupt mining, and impede confirmations.The same 51% vulnerability can arise in PoS when one miner owns more than half of the coins.
- Consensus and account security: Lost private keys cannot be recovered, while stolen keys allow account tampering and make criminal activity difficult to trace or reverse.Blockchain’s lack of a centralized trusted institution limits recovery after key theft.
- Illicit use: Bitcoin’s pseudonymous addresses and trading platforms facilitate ransomware, underground-market transactions, and money laundering.Examples include CTB-Locker ransomware, Silk Road, and privacy-enhancing tools such as Dark Wallet.
- Transaction integrity: Double spending remains possible despite transaction validation because race attacks exploit the interval between transaction initiation and confirmation.An attacker can obtain the first transaction’s output before the conflicting transaction is mined as invalid.
- Transaction privacy: Blockchain privacy protections can fail when Monero mixin sampling is absent or nonrandom, enabling transaction-input inference with 80% accuracy.66.09% of transactions had no mixins, and 62.32% of inputs with mixins were deducible.
3.2. Specific Risks to Blockchain 2.0
Blockchain 2.0 introduces smart-contract-specific risks, including criminal uses, programming vulnerabilities, inefficient code, and under-priced operations. Reported studies identify widespread vulnerable or costly contracts and motivate protocol-level gas revisions.
- Criminal smart contracts: Criminal smart contracts can support confidential-information leakage, cryptographic-key theft, and real-world crimes such as murder, arson, and terrorism.The PwdTheft example uses an SGX application to obtain and transmit encrypted password data for a potential reward.
- Smart-contract vulnerabilities: 8,833 out of 19,366 Ethereum smart contracts were found vulnerable to four security bugs identified with symbolic execution.The surveyed bug classes include transaction-ordering dependence, timestamp dependence, mishandled exceptions, and reentrancy.
- Under-optimized smart contracts: More than 80% of 4,240 real Ethereum smart contracts contained at least one of three gas-costly patterns detected by Gasper.The patterns are dead code, opaque predicates, and expensive operations inside loops.
- Under-priced operations: Under-priced operations can let attackers consume disproportionate computing resources and significantly slow block synchronization.An attacker called EXTCODESIZE more than 50,000 times in one transaction because its gas value was only 20.
- Under-priced operations: EIP150 modified the gas values of seven IO-heavy operations to address security problems caused by under-pricing.The revised gas table was implemented on Ethereum’s public chain after block 2,463,000.
4. Attack Cases
The paper surveys real attacks spanning consensus, smart contracts, network routing, peer connectivity, transaction liveness, and subgroup communication. These cases show how attackers exploit private-chain strategies, contract control flow, routing interception, isolation, delayed confirmation, and partitioned mining power.
- Consensus attacks: Selfish mining withholds discovered blocks to build a private fork, waste honest miners’ computations, and potentially obtain undue rewards.The Selfish-Mine strategy analyzes cases where public and private chains diverge or temporarily fork at equal length.
- Smart-contract attacks: The DAO attacker exploited reentrancy by repeatedly invoking a callback during withdrawals and stole about 60 million US$ after the contract raised 150 million US$.The attack occurred 20 days after deployment of the Ethereum crowdfunding contract.
- Network attacks: BGP hijacking rerouted Bitcoin miners’ connections to an attacker-controlled mining pool and collected an estimated 83,000 US$ over two months.Detection does not immediately resolve the attack because remediation may require hours of human-driven configuration changes.
- Network attacks: An eclipse attack monopolizes a victim’s incoming and outgoing connections, isolating the victim and filtering its blockchain view or redirecting computing power.The attack can force work on obsolete blockchain views and support further malicious activity.
- Transaction-liveness attacks: A liveness attack delays target-transaction confirmation through preparation, transaction denial, and blockchain-retarder phases.The attacker builds a private-chain advantage, withholds the target transaction’s block, and later publishes it while extending the private chain.
- Subgroup communication attacks: A balance attack can enable double spending with about 5% of total computing power by delaying communication between subgroups with similar mining power.The attack violates persistence of the main-branch prefix and can cause a merchant to accept goods before an alternative transaction prevails.
5. Security Enhancements
The paper reviews security enhancements addressing decentralization, performance-security tradeoffs, smart-contract bugs, privacy leakage, and off-chain data access. These solutions include decentralized mining pools, quantitative analysis, symbolic execution, privacy-preserving contracts, and authenticated data feeds.
- 5.1. SmartPool: SmartPool decentralizes mining-pool operation through a smart contract whose state is maintained by Ethereum rather than a pool operator.
- 5.1. SmartPool: Batch submission and partial share verification make SmartPool more efficient than a peer-to-peer pool.
- 5.1. SmartPool: SmartPool prevents share resubmission across batches and guarantees honest miners’ expected rewards despite malicious pool participants.
- 5.2. Quantitative Framework: A quantitative framework combines blockchain simulation with an MDP-based security model to analyze performance and attacks including double spending and selfish mining.The simulator produces statistics such as propagation time, block size, network delay, stale-block rate, and throughput; the security model outputs optimal adversarial strategies.
- 5.3. Oyente: Oyente uses symbolic execution of Ethereum smart-contract bytecode, following the EVM execution model to detect bugs in deployed contracts.Its process builds and enriches a control-flow graph before simulated execution using the Ethereum global state.
- 5.4. Hawk and 5.5. Town Crier: Hawk protects privacy by separating private and public contract portions, while Town Crier bridges HTTPS data sources and smart contracts through signed blockchain messages.Hawk uses an SGX-based manager, whereas Town Crier isolates network communication from execution across Ethereum, SGX, and HTTPS components.
6. Future Directions
The paper identifies future research directions based on its systematic security examination. It emphasizes more efficient consensus mechanisms, stronger privacy protections for feature-rich dAPPs, and improved blockchain-data management.
- More efficient consensus mechanisms are needed to address PoW’s waste of computing resources, including hybrid PoW–PoS approaches.
- Growing feature-rich dAPPs increase privacy-leakage risks, motivating techniques such as code obfuscation, application hardening, and trusted execution.
- Blockchain-data management remains a research direction because stored block information, transaction data, and contract bytecodes are not necessarily all valid.
7. Conclusion
The paper systematically examines blockchain security risks, real attacks, and security enhancements across popular systems such as Ethereum, Bitcoin, and Monero. It analyzes vulnerability causes and consequences and suggests future research directions.
- The study examines security risks in popular blockchain systems and analyzes the causes and possible consequences of each risk or vulnerability.
- The paper surveys real blockchain attacks and analyzes the vulnerabilities exploited in those attacks.
- It summarizes blockchain security enhancements and suggests future directions for research.