Key Takeaways:
- Aptos started hiding details of pending transactions in its Encrypted Mempool on mainnet.
- Validators are not the first to process transactions, thereby mitigating the possibility of frontrunning and sandwich attacks.
- The system brings in around 20 milliseconds of the delay per block for 50 nodes.
Aptos has activated a protocol-native encrypted mempool on mainnet, giving users an optional way to hide transaction details until block ordering is finalized. The feature aims at mitigating a longstanding issue in DeFi: frontrunning and sandwich attacks on visibly pending transactions via MEV tactics.
Table of Contents
Aptos Encrypts Transactions Before They Enter the Mempool
If the feature is enabled by the users, its transaction payload is encrypted on the client before being sent to the network. Sensitive information like function call and arguments are hidden, but the sender address and gas fee are still visible.
Validators then confirm the ordering of the transactions without having to view the payment message itself. Validators then collectively decrypt the transactions if consensus has agreed on the block, allowing them to execute the transactions and add them to the blockchain record.
No Single Validator Can Unlock a Pending Transaction
Aptos implements threshold encryption directly based on its current set of validators. No single validator is able to decode a pending transaction nor is any of the group below its required stake threshold.
We do not create any additional decryption committee or any private block builder or trusted hardware provider in Aptos. Rather than that, AQUA token conversion is handled by the same validators that facilitate the Aptos consensus.
Encrypted Mempool Targets Frontrunning and Sandwich Bots
The public mempools show transaction data prior to execution. That’s where automated trading bots find some opportunity to spot huge transactions, copy or execute trades before and after another trader.
With the payload secret and confirmed repayment order solved, Aptos eliminates numerous of the information these bots need to disrupt holding trades.
Protection may be particularly relevant for institutional orders, big DeFi trades and automated agents, as early disclosure of transaction intention may result in execution risk.
Petra is first to make the feature available as part of its signing process, letting users choose to encrypt a transaction at sign time. Additionally, a perpetual futures exchange on Aptos, Decibel, will be using encrypted mempool infrastructure to ensure that orders remain off-chain until they are executed.
Aptos Adds Privacy With About 20ms of Extra Latency
Running threshold encryption at blockchain scale adds to the computation required, but according to Aptos, the performance effect is minimal.
Its batched encryption design will allow it to decrypt an entire block with constant-size communication per validator, regardless of the number of transactions in the batch.
Testing with 50 nodes resulted in an additional 20 milliseconds per block over the production version of AptosBFT. The delay is insignificant as Aptos blocks are finalized in less than a second.
Restrictions in both of the first implementations prevent complete transaction anonymity. Sender addresses and gas information are still visible, while the transaction payload itself remains encrypted until block finalization.
Existing Move smart contracts do not require changes, and support is available through the Aptos TypeScript SDK version 7.0.0 and later.
