Cryptocurrency: Unlocking the chains

Blockchain is a digital technology deployed in 2008 with the emergence of Bitcoin. This cryptocurrency was designed in particular to address the loss of trust in financial and government institutions associated with the economic crisis. Although not exclusive, the main current applications of blockchains concern transactions associated with financial systems. "With blockchain, trust in institutions is replaced by trust in technology," explains Petr Kuznetsov, researcher at Télécom Paris and member of the TrustShare: blockchain-oriented innovation chair consortium. The principle behind this trust is to decentralize information using a peer-to-peer network. Instead of storing information on a single server, with operations controlled by a trusted third party, algorithms are used to distribute it between different Internet users, who can either be the server or the receiver of another.

The reliability of a blockchain depends on its robustness against accidental errors and intentional attacks, known as Byzantine attacks, aimed at falsifying information. The main security problem with cryptocurrencies relates to the timing of transaction validation throughout the system. Double-spending is when a participant carries out two transactions with the same money, taking advantage of the fact that the first transaction has not yet been taken into account by the system as a whole to carry out the second. Synchronization protocols are therefore put in place to guarantee the integrity of shared information. They use so-called consensus methods between participants, meaning that they must agree on an identical version of the blockchain. "This corresponds to the algorithmic principle of state machine replication, where consensus must be resolved for each block of information added and each participant has access to an identical copy of the blockchain" points out researcher Petr Kuznetsov.

Numerous consensus-building methods have been introduced over the years. By way of example, the first existing method used in the case of Bitcoin is called proof of work. With this approach, the addition of a block requires the solution of very complex cryptographic problems, involving considerable computing power. This system aims to slow down the addition of new blocks in order to guarantee the synchronicity of the information available to participants. This prevents a dishonest user from double-spending before the blockchain is updated for all users. "For this validation to work, more than 50% of the computing power must belong to correct participants. This helps combat attacks: if a minority user adds or modifies a block fraudulently, then it will not be validated by the consensus," explains the researcher. These technological and algorithmic constraints make attacks difficult. Other consensus models work with different proof-of-trust designs, but on the whole these models are not always optimized.

"Consensus-based solutions work, but they remain extremely complex, cumbersome and energy-intensive, which can be excessive and unnecessary in some cases," says the researcher. "We are looking to develop solutions that are less cumbersome and require less power to find more suitable solutions," he continues.

Streamlining computing protocols

In order to design more optimized protection systems, the research carried out by the TrustShare: blockchain oriented innovation chair consortium consists in hypothesizing and designing models of attackers, based in particular on the degree of trust between participants in the system. This depends on the ease of access to the blockchain, and therefore on the openness of the environment. Environments open to all participants, such as the Bitcoin environment, have a higher probability of containing malicious participants than environments where access is controlled. Open environments are those with the most uncertain participants, and require more expensive consensus resolution models.

A first approach to lightening the load on calculations is to improve trust in the environment by developing models that take into account more localized spaces within blockchains themselves. In this case, despite the fact that the degree of trust in the general system is not high in the system as a whole, it can be higher in subsets. "Trust in cryptocurrencies is like the financial market, where it is higher in national markets because regulated by national laws," explains the researcher. Thus, it would be possible to implement less cumbersome transaction validations, which would theoretically have weaker security guarantees in the global system but work perfectly in a more local space. "Local trust assumptions enable lighter, more efficient solutions" points out the researcher. With the arrival of Bitcoin, blockchain innovation has based its operation on addressing very complicated calculations in hostile environments where there are no models incorporating trust assumptions. Although functional, it would be cost-effective to strengthen these models by adding assumptions, as well as to reduce certain non-optimal calculations.

The study of the structure of blockchains in terms of the participants and the links between them also enables certain mechanisms to be streamlined. For example, it would be possible to dispense with the system of consensus on the absolute order of blocks, and instead rely on partial orders. Indeed, between different parts of a blockchain, certain transactions sometimes have no influence on each other, and it is not necessary to determine which transaction takes place first in relation to another. "In the end, it's quite rare for the integrity of cryptocurrency transactions to actually depend on the total order of the blocks of information," the researcher points out. Consensus on an absolute order is necessary when what is implemented in a system gives rise to potential conflicts between different parties. But in many cases, certain blocks can be arbitrarily ordered without affecting the system. "Just like a human society: each individual action does not depend on all the others, some may have more influence on the system as a whole, but most local and ordinary actions have no immediate consequence for others" illustrates the researcher.

Current research in this field aims to identify cases where conflicts may arise and where it is absolutely necessary to maintain a consensus, as opposed to cases where this is not necessary. The consensus method, which is seen as the main way of solving cryptocurrency security issues, has significant disadvantages in terms of non-optimized expenditure of energy and time. With a finer understanding of the blockchain environment, these methods could be adjusted on a case-by-case basis.

The TrustShare: blockchain-oriented innovation chair project is being developed at Télécom Paris, part of the Carnot Télécom & Société numérique, and is funded by Mazars Conseil and the Caisse des Dépôts et Consignations. "We are aggregating data on a large scale in order to solve these highly complex problems, which have major implications for the future of decentralized information systems," concludes the researcher.