Research
zkSAS: Practical Zero-Knowledge Proofs for Verifiable Spectrum Access Management
Today's CBRS Spectrum Access Systems are centralized: commercial users must blindly trust that the operator ran correct interference calculations and complied with FCC protection rules, and there is…

- arXiv
- 2609.28699
- Published
- 2026-09-23
- Authors
- Nishat F. Purbasha, Ifteher Alom, Eric W. Burger, Y. Thomas Hou, Wenjing Lou, Yang Xiao
AI summary
Overview
Research area: Applied cryptography for wireless spectrum policy — specifically zero-knowledge proof (ZKP) systems built for Dynamic Spectrum Access (DSA) and Spectrum Access System (SAS) deployments, with a secondary focus on blockchain-based spectrum management.
Technical level: Advanced. The paper assumes familiarity with zk-SNARKs (Groth16), arithmetic circuits and R1CS, Circom, and CBRS regulatory concepts such as PAL/GAA tiers, DPAs, and the move list. The prose is readable, but the core content is constraint arithmetization.
Scope (one sentence): The paper designs, implements, and benchmarks a suite of Groth16 zero-knowledge circuits that prove a spectrum allocation — and a move-list suspension decision — satisfies CBRS regulatory constraints without revealing the underlying user data.
What This Paper Is About
Today's CBRS Spectrum Access Systems are centralized: commercial users must blindly trust that the operator ran correct interference calculations and complied with FCC protection rules, and there is no way to audit a closed server. Blockchain-based alternatives make allocation transparent but expose sensitive user data (device IDs, geolocation, access patterns) on a public ledger and make rich allocation algorithms computationally prohibitive on-chain. zkSAS's goal is to let an SAS operator prove that its channel assignments and incumbent-protection decisions obey regulatory constraints, while keeping the assignment data private and keeping verification cheap enough to run on a client or a smart contract.
Key Contributions
-
A ZKP circuit suite for allocation constraint validity. Seven circuits (compiled with Circom into R1CS) that arithmetize the allocation constraints from prior work — PAL assignment per channel-and-county, PAL assignment per county, PAL interference protection, GAA assignment, GAA mutual interference control, and incumbent interference protection — plus a seventh added "allocation integrity" constraint that binds the private assignment witness to a public hash commitment.
-
A commit-and-prove design using Poseidon. The channel assignment indicators (x, y, Z), license counts, RSS values, distances, and thresholds are all private witnesses; only a Poseidon hash commitment over the flattened allocation plus instance metadata is public. Poseidon is chosen over SHA-256 because its arithmetic-native structure drastically reduces the number of R1CS constraints.
-
A ZKP construction for move-list validity. Rather than encoding the full standardized move-list algorithm in-circuit, the authors prove the validity of its critical checks: that a supplied list is correctly sorted, that the cutoff index is the largest satisfying the 95th-percentile aggregate interference threshold, and — in a personalized proof per suspended user — that the user's grant identifier falls after that cutoff.
-
Three use cases and a prototype evaluation. The paper shows zkSAS as an audit tool for centralized SAS, as an off-chain execution layer for blockchain-based SAS (using BD-SAS and
FinishAlloc/FinishSuspendcontract functions), and as a possible enabler of "Spectrum Bux" payments; it then benchmarks the prototype on a laptop and on Hyperledger Fabric.
Main Findings
- Proof generation is the dominant cost; verification is cheap. Across both the centralized and blockchain settings, the authors report that proof generation dominates total overhead, while verification is efficient and largely insensitive to the spectrum user population.
- Scaling depends on the constraint. Proof generation grows with the encoded county count and user population, but the growth rate varies widely: Constraints 2 and 4 stay nearly constant as GAA users per county increases, while Constraint 5 (
O(A·N²), the pairwise GAA distance check) shows the steepest increase and becomes the dominant contributor at larger county and user counts. - The commitment circuit often dominates end-to-end proving time. Although Constraint 7 scales only linearly in the flattened witness length
L, its large per-permutation constant means end-to-end proving time is frequently close to the Constraint 7 proving time. - Verification is constant-time per proof. Groth16 verification is independent of circuit size — a constant number of pairings plus a small linear term in public inputs — so total verification scales with the number of proofs (one bundle of seven), not their complexity. The paper's complexity table gives per-constraint verification as
O(F)for Constraints 1, 3, 5, 6, 7 andO(A)for Constraints 2 and 4. - Generation complexity per constraint (proof gen / proof file size): Constraint 1
O(A)/O(1); Constraint 2O(A)/O(1); Constraint 3O(A²·N)/O(1); Constraint 4O(A·N)/O(1); Constraint 5O(A·N²)/O(1); Constraint 6O(A·N)/O(1); Constraint 7O(L)/O(1). - Move-list proofs are linear in prover time, constant in verification. For a protection point and channel with grant set size
G, proof generation isO(G)and verification isO(1). - Parallelization helps. Proving is split into a serial phase (compilation, trusted setup, witness generation) and a parallel phase (one Groth16 prover process per constraint), so wall-clock proving time is set by the slowest constraint prover rather than the sum of all provers.
- Constraint 6 is optional. When no incumbent is active during an allocation computation, the allocation proof does not need to include Constraint 6, reducing overhead.
- Latency is described as practical at realistic county- and user-scale workloads. The specific millisecond and second timings are in the evaluation figures, which are not present in the excerpt reviewed here, so no numerical latency values are reported in this summary.
Methodology in Plain English
The researchers reframed regulatory compliance as a mathematical statement that can be proven without being shown. Each SAS rule (for example, "at most one PAL user per channel per county," or "aggregate interference at this protection point stays under threshold T") is translated into an arithmetic circuit using Circom, a language for building the constraint systems that zk-SNARKs operate on. The actual allocation — who got which channel, the signal strengths, the distances, the thresholds — is fed in as the prover's secret witness, so nothing about individual users leaks to the verifier.
Because the allocation itself is hidden, a dishonest operator could show one allocation to the proof and a different one to users. To close that hole, the authors add a commitment: a Poseidon hash of the flattened assignment vector plus metadata (county count, user counts, channel index) is published as the only public signal, and the circuit enforces that the private witness hashes to it. Users receive the real allocation, the auditor sees only the hash and the proofs, and all seven proofs must agree on the same commitment.
Real-valued quantities are handled with decimal fixed-point encoding (store ⌊x·10^f⌉), plus bit-length range checks on inputs and intermediate sums so modular wraparound in the prime field cannot accidentally satisfy an inequality. For the move list, the authors deliberately avoid implementing sorting and interference aggregation in-circuit: those run off-chain and their outputs — the sorted list and the cutoff index — are supplied as witnesses, and the circuit merely proves the list is nondecreasing, that the cutoff is the largest index meeting the 95th-percentile threshold, and that the user's grant ID sits after the cutoff. This yields one small personalized proof per suspended user instead of one enormous proof for the whole move list.
The prototype is roughly 730 lines of Circom using circom v2.1.x and snarkjs with Groth16, tested on a MacBook Pro with an Apple M4 Pro and 24 GB of memory, with on-chain verification tested on a locally deployed Hyperledger Fabric network running on Ubuntu 20.04 LTS with an Intel Xeon E5-2630 v3 at 2.40 GHz and 32 logical processors.
Why This Matters
Impact on research. zkSAS reframes spectrum policy compliance as a provable statement rather than a trusted assertion, and it does so at the granularity of real regulatory constraints (CBRS PAL/GAA rules, DPA incumbent protection, the standardized move list) rather than toy examples. It also demonstrates a pattern for the broader "verifiable off-chain computation" literature: complex algorithmic steps like sorting and percentile-based aggregation can stay off-circuit, with proofs covering only the checks that matter. The commit-and-prove binding of an allocation to a single hash is a reusable answer to the witness-substitution problem that arises whenever an operator is both prover and data owner.
Real-world applications:
- Regulatory auditing of commercial SAS deployments. An external auditor or the FCC could verify that a nightly CPAS allocation satisfied interference and licensing constraints without the operator disclosing proprietary algorithms or user records.
- Privacy-preserving incumbent protection. A naval radar or satellite ground station arriving in a DPA could receive a verifiable, personalized proof that affected CBSDs were correctly identified for the five-minute suspension, without exposing those users' identities or locations.
- Blockchain-based spectrum marketplaces. Off-chain allocation with on-chain proof verification lets a spectrum access contract release payment (
FinishAlloc) or refund unused deposits (FinishSuspend) only on verified compliance, sidestepping both ledger transparency and replicated-execution cost. - Anonymous or sensitive-user access. The Spectrum Bux model sketched in the paper allows a user — for instance a military user — to prove sufficient balance for service without revealing identity or holdings, with settlement deferred.
Industry relevance. The design targets both the installed CBRS SAS model (3.55–3.7 GHz, deployed since 2020) and emerging decentralized spectrum markets, so it does not require an operator to abandon its current architecture. Because verification is constant-time per proof, it is viable in client devices and in smart contracts, and because allocation algorithms remain off-chain and private, SAS vendors do not have to publish the intellectual property embodied in their allocators.
Future Directions
- Reduce Constraint 5's
O(A·N²)cost, which the authors identify as the steepest-growing and eventually dominant proving cost; the paper does not report a mitigation. - Develop the Spectrum Bux extensions, including user-side zero-knowledge balance proofs and deferred financial settlement with band managers, which the authors explicitly defer to future work.
- Narrow the full end-to-end latency picture at larger county and user scales, and characterize how the proof bundle behaves when the number of constraints or the commitment length
Lgrows — the excerpt reviewed here covers constraint-wise and end-to-end timings at Total County = 13. - Integrate proven allocations more tightly with on-chain state, for example generalizing the
FinishAlloc/FinishSuspendpattern beyond the BD-SAS example to other blockchain SAS designs and to trading or short-term leasing logic.
Target Audience
Researchers and practitioners working at the intersection of applied cryptography and wireless policy: zk-SNARK circuit designers looking for a realistic, constraint-dense application; SAS administrators, spectrum regulators, and spectrum-policy researchers evaluating auditability mechanisms for CBRS and successor frameworks; and blockchain/Web3 engineers building verifiable off-chain computation for physical-infrastructure markets. Readers without a cryptography background will find the problem framing and use cases accessible, but the constraint arithmetization and complexity analysis sections require comfort with R1CS and zk-SNARK internals.
Authors’ abstract
Dynamic Spectrum Access (DSA) through the Spectrum Access Systems (SAS) elevates spectral efficiency, yet existing centralized models face allocation logic opaqueness and a lack of independent verifiability. While blockchain-based SAS architectures offer transparency and verifiability by default, they introduce critical privacy risks and prohibitive on-chain computational overhead. We introduce zkSAS, a practical zero-knowledge proof (ZKP) system designed to address the…
The full abstract on arXiv