Skip to content
Crypto.world

The Ledger

GRIDSHIELD Preprint Proposes Order-Independent Clearing for P2P Energy Trading

A preprint proposes a system that combines sealed-bid, order-independent clearing with proofs that allocations satisfy public linear grid constraints. Its evaluation reports a few-kilobyte feasibility proof, no trusted setup and throughput of about 1,064 transactions per second.

GRIDSHIELD Preprint Proposes Order-Independent Clearing for P2P Energy Trading

GRIDSHIELD reports about 1,064 transactions per second in a live-network benchmark.

A preprint submitted to arXiv on 4 Nov 2026 describes GRIDSHIELD, a system for peer-to-peer energy trading that aims to make market clearing auditable without revealing private allocations. Its authors, Abdullahil Kafi, Md Zahidul Islam and Alvi Ataur Khalil, report that the system sustained approximately 1,064 transactions per second on a live Hyperledger Fabric network benchmarked with Hyperledger Caliper. The evaluation also reports a few-kilobyte proof for an entire round and no trusted setup. In tests against three prior designs, GRIDSHIELD rejected infeasible allocations that two of them accepted.

The preprint opens with the concern that “Distributed energy resources, rooftop photovoltaics, home batteries, and controllable loads, are turning consumers into prosumers who both produce and consume power.” Local markets let those participants trade surplus energy directly, but energy exchanges must account for physical limits on the shared distribution grid. The paper identifies thermal ratings and voltage bounds as examples of constraints that a valid market outcome must respect.

The design addresses the clearing operator’s power to order trades.

In the paper’s description of a permissioned consortium, the distribution system operator typically operates the ordering service and clears the market. The ordering service sequences transactions before they are added to the blockchain. The operator therefore has influence over which participants are prioritized, who is curtailed when the feeder is congested, and who pays congestion charges.

The authors argue that this creates an opportunity for manipulation if the operator is compromised or coerced. It could reorder trades for profit or curtail a rival, while participants whose bids and consumption data are hidden may have difficulty distinguishing a fair outcome from favoritism. The paper contrasts this problem with earlier approaches: network-aware clearing can respect grid constraints while operating on plaintext allocations, while privacy-preserving designs may omit network constraints, enforce them inside an unauditable computation, or require a per-circuit trusted setup.

GRIDSHIELD clears sealed bids with an order-independent batch program. The paper says the outcome depends on the bid set rather than the order in which transactions arrive, removing the welfare a compromised operator could extract by reordering trades. It then uses an aggregated congestion-feasibility proof, or ACFP, to let others verify that the hidden allocation satisfies the supported grid constraints.

ACFP is described for public linear constraints.

The proof uses the additive homomorphism of Pedersen commitments. GRIDSHIELD reconstructs each supported constraint from commitments associated with individual buses, then reduces feeder-wide feasibility checks to a single aggregated range proof over the hidden allocation. The paper describes this method for public linear constraints, including linearized grid constraints; it does not establish that ACFP handles arbitrary nonlinear limits. The authors also describe a reveal-and-recompute challenge intended to catch clearings that are feasible but unfair.

The preprint introduces congestion-responsibility fairness, or CRF, as a metric for assessing curtailment. Rather than measuring fairness against consumption alone, CRF scores curtailment against each party’s physical responsibility for congestion.

The evaluation covers four feeders.

The reported comparison includes three prior peer-to-peer energy-trading designs and four feeders, one of which is a 240-node Midwest feeder using real hourly loads. The system was benchmarked on a live Hyperledger Fabric network. The authors report that GRIDSHIELD proves a full round feasible with a few-kilobyte proof and no trusted setup, while rejecting infeasible allocations accepted by two of the three comparison designs.

Sources

  1. Original source · Preprint arXiv arXiv:2611.05816

Related stories

Explore

More articles