# Ethereum IoT preprint finds batched ECDSA costs 8,963 gas per record

Published 2026-11-19 · Crypto.world Team · Crypto.world

> A new arXiv study measures Ethereum’s costs for authenticating and anchoring IoT records after Pectra and Fusaka. It reports that batched secp256k1 ECDSA costs 8,963 gas per record, while BLS aggregation costs more and fits fewer records in one transaction.

## Batched signatures lead on cost

Batching IoT records and anchoring them with a Merkle tree brought Ethereum authentication costs to 8,963 gas per record in a new [arXiv](https://arxiv.org/abs/2611.23764) preprint. The paper, submitted on Nov. 18 and not yet peer-reviewed as far as the arXiv page shows, measures signature and anchoring designs under rules before and after Ethereum’s Pectra and Fusaka upgrades.

The researchers compare four replay-protected signature designs: ECDSA over secp256k1 and P-256, BLS aggregate signatures, and a pseudonymous identity-based signature with batch verification. Their tests use signed transactions and report per-record gas costs, transaction capacity and signing times alongside different levels of trust in an aggregator.

In the study’s tests, batching and Merkle anchoring cut the cost of ECDSA authentication sixfold, to 8,963 gas per record. At 500 records, the paper reports that figure for secp256k1 ECDSA, compared with 14,542 gas per record for P-256.

By contrast, the preprint’s table gives the identity-based design (pseudonym-grouped batching) 24,634 gas per record at 500 records and BLS aggregation 34,927 gas per record at 100 records; BLS batches in its tests top out at 411 records. The paper attributes BLS’s higher on-chain cost largely to hashing messages to the curve, even though aggregation reduces how many signatures must be included in transaction data.

## Transaction limits shape batch sizes

The study’s comparison of transaction capacity puts a limit on the apparent benefits of batching: one transaction can carry up to 1,754 ECDSA-authenticated records, but 411 BLS-aggregated records. The paper says the Fusaka transaction gas cap is the constraint for larger batches in designs whose costs grow with batch size.

For device-side signing on a Raspberry Pi Zero 2 W, the authors measured pairing-based signing at 5.7 milliseconds, against 0.29 milliseconds for ECDSA. They say pairing-based designs may be worthwhile when an application needs their additional properties, rather than for lower on-chain cost.

## Three levels of trust

Under the study’s three-level framework, T0 has a trusted aggregator verify records off-chain and post a Merkle root; T1 posts record hashes; and T2 has a contract verify each device signature before anchoring records. The researchers say T2 allows records to be authenticated even if the aggregator is malicious, while the lower-trust levels rely more heavily on the aggregator’s account of authenticity.

The paper describes a supply-chain finance scenario in which camera units photograph crops, store images off-chain and upload signed fingerprints. A consortium-chain aggregator collects and verifies records, while Ethereum is used to anchor them; in the proposed workflow, production records can support checks at financing milestones.

## What the upgrades changed

The authors say Pectra and Fusaka mainly changed which designs could be implemented on-chain, rather than the costs of designs available under multiple rule sets. Pectra added BLS12-381 precompiles, while Fusaka added a P-256 verification precompile; the study also examines Fusaka’s transaction gas cap and changes to MODEXP pricing.

The study notes trade-offs beyond gas: it says the identity-based design involves key escrow at the key generation centre. Its limitations include a synthetic workload with 10 devices per batch and no energy measurements; it adds that costs on rollups would depend on their data-availability pricing.

## Sources

1. Original source · Preprint [Authenticating IoT Data on Ethereum after Pectra and Fusaka: A Cost Study of Signature Schemes, Batching and Anchoring for Agricultural Supply Chain Finance](https://arxiv.org/abs/2611.23764) arXiv:2611.23764

## Related stories

- [Preprint: ML-DSA-44 Per-Record Design Costs 8.7× ECDSA Even With Proposed Precompile](https://crypto.world/news/ethereum-ml-dsa-record-signing-calldata-cost)
- [Preprint proposes fee-memory rule for Layer 2 posting costs](https://crypto.world/news/preprint-layer-2-dynamic-fees-vault-approach)
- [Preprint proposes one-transaction proofs for Ethereum quantum migration](https://crypto.world/news/researchers-propose-one-transaction-proofs-ethereum-quantum-migration)
- [Ethereum developers weigh Glamsterdam spec changes that could push activation to May 2027](https://crypto.world/news/ethereum-glamsterdam-upgrade-could-slip-to-may-2027)
- [Ethlabs proposes Ethereum post-quantum readiness in 2027](https://crypto.world/news/ethlabs-ethereum-post-quantum-readiness-2027)

## Explore

[More articles](https://crypto.world/news)

---

Canonical: https://crypto.world/news/ethereum-iot-authentication-cost-study-pectra-fusaka
