B5GRoam: Zero-Trust On-Chain Roaming Settlement
- B5GRoam is a zero-trust roaming settlement framework leveraging zkSNARK proofs for cryptographically verifiable billing reconciliation between mobile operators.
- It employs smart contracts and Layer 2 zk-rollups to enable efficient, scalable, and trustless reconciliation of call detail records without revealing sensitive data.
- B5GRoam extends into 5G-and-Beyond robotics and autonomous mobility by integrating communication-aware path planning, localization, and dynamic multi-domain orchestration.
Searching arXiv for the named topic and closely related papers. arxiv_search({"query":"B5GRoam roaming blockchain zero trust zkSNARKs OR \"B5GRoam\" robotics autonomous mobility", "max_results": 10, "sort_by": "relevance"}) B5GRoam is a label used in the supplied literature in two closely related Beyond 5G (B5G) contexts. In the most explicit usage, it denotes a “zero trust, decentralized framework” for “secure, efficient, and privacy-preserving on-chain roaming settlements” between mobile network operators, with cryptographically verifiable call detail record submission, zkSNARK-based privacy preservation, and Layer 2 zk-Rollups for scalability (Rezazi et al., 19 Sep 2025). In a broader contextual usage, the term is also associated with “5G-and-Beyond Robotics and Autonomous Mobility,” where communication-aware robotics, localization, and autonomous mobility are integrated with B5G networking (Liu et al., 2020). Across these usages, B5GRoam is situated at the intersection of roaming, mobility, verifiability, privacy, and dynamic multi-domain orchestration in B5G systems (Rezazi et al., 19 Sep 2025, Raisanen et al., 2020).
1. Terminological scope and conceptual placement
In the roaming-settlement literature, B5GRoam is introduced as a framework addressing “secure, efficient, and trustworthy mechanisms for billing reconciliation between mobile operators” in 5G and beyond networks (Rezazi et al., 19 Sep 2025). Its explicit target is roaming settlement, particularly the long-standing dependence on centralized intermediaries such as clearing houses, and the associated issues of “data privacy risks,” “assumptions of mutual trust,” and “scalability bottlenecks” in blockchain-based or legacy solutions (Rezazi et al., 19 Sep 2025). The framework therefore belongs to the class of decentralized inter-operator settlement architectures, but distinguishes itself by combining on-chain enforcement with zero-knowledge verification and rollup-based throughput scaling (Rezazi et al., 19 Sep 2025).
The broader B5G literature places roaming within a larger cross-domain service environment. “Cross-stakeholder service orchestration” generalizes 5G network slices by enabling domain operators to expose functionalities as “capabilities” associated with metadata and Service Level Agreements (SLAs), which are then composed into end-to-end services across multiple stakeholders (Raisanen et al., 2020). That literature states that such a framework “generalizes and extends traditional roaming,” replacing rigid bilateral agreements with “dynamic capability exposure” and automated cross-domain selection (Raisanen et al., 2020). This suggests that B5GRoam can be understood not only as a settlement primitive, but also as part of a wider B5G transition from bilateral roaming to programmable, service-driven, and SLA-governed inter-domain coordination.
A further contextual use appears in robotic communications. There, the supplied summary explicitly refers to a “B5GRoam (5G-and-Beyond Robotics and Autonomous Mobility)” setting characterized by communication-aware path planning, integrated localization and radio mapping, and real-time adaptation to radio and physical environments (Liu et al., 2020). A plausible implication is that the label B5GRoam is not yet lexically uniform across subfields: in one strand it names an on-chain roaming framework, while in another it functions as a robotics-and-autonomous-mobility shorthand (Rezazi et al., 19 Sep 2025, Liu et al., 2020).
2. Zero-trust roaming settlement architecture
The architectural core of B5GRoam consists of three primary actors: the User Equipment (UE), the Home MNO (HMNO), and the Visited MNO (VMNO) (Rezazi et al., 19 Sep 2025). The UE is assumed to contain a Trusted Execution Environment, such as “ARM TrustZone,” for secure handling of sensitive data and cryptographic operations (Rezazi et al., 19 Sep 2025). The HMNO is the payer in settlements, while the VMNO is the payee providing roaming services (Rezazi et al., 19 Sep 2025).
The framework enforces zero trust by replacing bilateral confidence with “cryptographic verifiability” and “programmatic settlement enforced by on-chain logic” (Rezazi et al., 19 Sep 2025). Smart contracts encode “all agreements, unit rates, and settlement logic,” so that neither the VMNO’s claimed usage nor the HMNO’s willingness to pay is accepted without proof (Rezazi et al., 19 Sep 2025). Concretely, the zero-trust model means that “VMNOs cannot unilaterally inflate or forge CDR data,” because only records reconciled with a user-side commitment are accepted, and “HMNO cannot repudiate charges that are cryptographically proven according to contractual terms” (Rezazi et al., 19 Sep 2025).
This model differs materially from roaming abstractions in cross-stakeholder orchestration. Capability-based orchestration exposes domain functions upward to an end-to-end service orchestrator and downward to domain orchestration and intent-based business management, with possible “SLA assurance, resource negotiation, and dynamic pricing” (Raisanen et al., 2020). B5GRoam, by contrast, is narrower and more operationally specific: it focuses on the settlement layer, where proof of service and proof of payment are mediated by on-chain logic rather than by marketplace discovery or intent translation (Rezazi et al., 19 Sep 2025). The two are nevertheless complementary. A plausible implication is that capability marketplaces and on-chain settlement could form adjacent layers in a future roaming stack: one for discovery and orchestration, the other for reconciliation and payment (Raisanen et al., 2020, Rezazi et al., 19 Sep 2025).
3. Verifiable CDR submission and proof constraints
A defining mechanism in B5GRoam is the “cryptographically verifiable CDR submission protocol” (Rezazi et al., 19 Sep 2025). When roaming starts, the UE’s TEE securely records local usage statistics denoted as , and computes a commitment hash using the Poseidon hash function:
This commitment is submitted to the blockchain (Rezazi et al., 19 Sep 2025). After the roaming session, the VMNO collects its own view of the same CDR data and must generate a proof showing that its usage record matches the previously posted user-side commitment and that the computed fee is consistent with the contract’s rates (Rezazi et al., 19 Sep 2025).
The proof circuit encodes two public verifiability constraints. The first is the settlement total:
The second is commitment consistency:
Here, the actual usage amounts remain private inputs, while the total fee and the hash commitment are public inputs visible on-chain (Rezazi et al., 19 Sep 2025). The protocol therefore establishes that the claimed amount is contractually correct and commitment-consistent without revealing granular CDR contents (Rezazi et al., 19 Sep 2025).
The paper’s algorithmic summary states that the workflow is: compute the billing total and hash check, assert both constraints in a zkSNARK circuit, prove the existence of private inputs matching the public inputs, and submit to the smart contract for verification (Rezazi et al., 19 Sep 2025). The significance of this design is that no single stakeholder has unilateral authority over the record of consumption. Settlement occurs only if independently observed data reconcile under the proof system (Rezazi et al., 19 Sep 2025).
4. Privacy model and scalability mechanisms
B5GRoam uses “Non-interactive Zero-Knowledge Proofs (zkSNARKs)” to make billing verification both trustless and privacy-preserving (Rezazi et al., 19 Sep 2025). The private inputs are the usage amounts, while the public inputs are the total fee and the precommitted hash (Rezazi et al., 19 Sep 2025). The system therefore prevents disclosure of “identity, timestamps, locations,” or other granular CDR data during verification (Rezazi et al., 19 Sep 2025). The supplied summary states that this “eliminates on-chain privacy leaks” and allows billing logic to be enforced without exposing user or operator activity (Rezazi et al., 19 Sep 2025).
Scalability is addressed through “Layer 2 (L2) solutions—specifically zk-Rollups (e.g., zkSync)” (Rezazi et al., 19 Sep 2025). Multiple settlement proofs are aggregated off-chain, and a single proof attests to the validity of a large batch (Rezazi et al., 19 Sep 2025). Only minimal data, such as the batch proof and hash, are verified on Layer 1 (Rezazi et al., 19 Sep 2025). This directly targets the throughput mismatch between 5G/B5G roaming workloads and ordinary Layer 1 blockchain capacity.
The evaluation reported in the supplied summary gives concrete scale indicators. For a 60-transaction batch, L2 Total Gas is “388,749” versus “15,636,180” for the L1 baseline, with savings “>96%”; for a 500-transaction batch, L2 Total Gas is “2,737,983” versus “130,301,500” on L1, with savings “>97%” (Rezazi et al., 19 Sep 2025). The framework reports “a throughput of over 7,200 tx/s,” and the evaluation section states “7,200 tx/s with 60-tx batch size” (Rezazi et al., 19 Sep 2025). The proving backend selected as optimal is “Groth16,” described as having “the lowest on-chain gas usage” and “stable subsecond latency/constant time verification” (Rezazi et al., 19 Sep 2025). The same summary reports “0.40 s/proof” and “~230,000 gas” for Groth16, while “UltraHonk” is reported as “<0.2 s” proving time but with “far greater verification gas costs” (Rezazi et al., 19 Sep 2025).
These figures should be interpreted in the narrow context of the reported experimental setup. The key architectural point is not simply that L2 is faster than L1, but that B5GRoam couples verifiable settlement and privacy preservation with a batching mechanism compatible with high-throughput mobile-network operation (Rezazi et al., 19 Sep 2025).
5. Relation to roaming privacy, orchestration, and mobility-aware networking
B5GRoam addresses one part of roaming privacy: billing and settlement confidentiality. Other supplied works address privacy and roaming from different layers. The “Virtual Private Mobile Network” model for B5G location privacy considers a set of UEs communicating in a D2D fashion and connected to the cellular network through multiple gateways, with the explicit purpose “to hide the position of the VPMN UEs to the mobile network operator” (Tomasin et al., 2021). That work derives a routing algorithm maximizing VPMN rate while imposing “for each UE the same data rate at each gateway” (Tomasin et al., 2021). It therefore protects location privacy by traffic equalization rather than by zero-knowledge settlement. A plausible implication is that B5GRoam and VPMN-type architectures address different observability surfaces: the former protects CDR verification, while the latter obscures gateway-usage patterns that could reveal physical position (Rezazi et al., 19 Sep 2025, Tomasin et al., 2021).
Roaming in B5G also appears in orchestration literature as a capability-composition problem rather than merely a charging problem. Capability orchestration exposes connectivity, edge computing, or AI/ML processing as discoverable, SLA-governed capabilities, and end-to-end service orchestration composes them across domains (Raisanen et al., 2020). The paper explicitly notes that this “generalizes and extends traditional roaming,” allowing not only connectivity but also “processing, computation, or AI access while ‘roaming’” (Raisanen et al., 2020). In that perspective, settlement frameworks such as B5GRoam can be interpreted as the accountability substrate for a more dynamic, multi-provider roaming economy.
Mobility-aware analysis adds another layer of context. In dense and heterogeneous 5G/B5G deployments, mobility affects “handoff rate, handoff probability, sojourn time,” and “users’ throughput or coverage” (Tabassum et al., 2018). The tutorial on mobility-aware analysis emphasizes that accurate roaming-related performance characterization is especially difficult in random, multi-tier, and high-mobility settings such as mmWave, vehicular, UAV, and high-speed rail scenarios (Tabassum et al., 2018). This suggests that a complete B5GRoam stack would require not only secure settlement, but also mobility-sensitive service assurance and handoff-aware accounting semantics.
6. Broader B5GRoam interpretations and adjacent research directions
The supplied material also associates B5GRoam with robotics and autonomous mobility. In “Robotic Communications for 5G and Beyond,” the summary states that the entire framework of signal and spatial models, simultaneous localization and radio mapping (SLARM), and intelligent trajectory and resource allocation is “specifically tailored for B5GRoam (5G-and-Beyond Robotics and Autonomous Mobility)” (Liu et al., 2020). In that usage, B5GRoam characterizes robotic systems operating in “high-stakes, dynamic, connected environments,” with “communication-aware path planning” and “real-time adaptation to both radio and physical environments” (Liu et al., 2020).
The technical elements of that strand differ from on-chain settlement but are conceptually linked by mobility and cross-domain service dependence. The paper introduces SLARM as a joint framework for building a geographic map and a radio map simultaneously, combining odometry, LIDAR or visual data, and real-time channel measurements (Liu et al., 2020). It also highlights RL-based trajectory and resource allocation for outdoor and indoor robots, with transfer learning for indoor multi-objective optimization (Liu et al., 2020). This suggests that, in some literature, B5GRoam is used to name an application domain in which roaming is not only inter-operator billing but also dynamic access to communication, localization, and computation resources during autonomous mobility.
Other adjacent B5G studies reinforce this broader interpretation. Intelligent backhaul selection uses network slicing, Integrated Access and Backhaul, satellite connectivity, and a DDQN agent to select backhaul links per slice under congestion, meeting slice-specific throughput and latency requirements (Morgado et al., 15 Jan 2025). Cloud-network integrated resource allocation jointly optimizes VNF placement, assignment, traffic prioritization, and path selection under capacity and delay constraints, with B{B}-CCRA and WF-CCRA as solution methods (Shokrnezhad et al., 2023). These are not B5GRoam papers in the narrow titular sense, but they show the infrastructural conditions under which dynamic roaming, mobility, and inter-domain service continuity become technically meaningful.
A common misconception would be to treat B5GRoam solely as a blockchain settlement mechanism or solely as a robotics-mobility paradigm. The supplied literature supports a more nuanced interpretation: the term has a precise use as the title of a zero-trust roaming-settlement framework (Rezazi et al., 19 Sep 2025), while also appearing as a broader B5G mobility label in robotic communications (Liu et al., 2020). The coexistence of these usages suggests an evolving terminology rather than a settled singular definition.
7. Limitations, trade-offs, and future development
The most explicit limitations in the B5GRoam settlement framework arise from the very properties it seeks to guarantee. Privacy preservation depends on zkSNARK construction and proof verification, while scalability depends on rollup aggregation (Rezazi et al., 19 Sep 2025). The paper positions Groth16 as the optimal backend because of its low on-chain gas usage and stable verification latency, but the reported comparison with UltraHonk indicates that proving and verification costs do not improve uniformly across all proof systems (Rezazi et al., 19 Sep 2025). This points to an implementation trade-off between prover performance and verifier efficiency.
At the system level, the framework is also scoped to roaming settlement rather than the full inter-operator service lifecycle. Capability-provisioning work indicates that future B5G roaming may involve dynamic discovery, SLA negotiation, intent-based management, and capability marketplaces (Raisanen et al., 2020). B5GRoam, in its current form, addresses the settlement trust problem after or during service consumption, but does not itself provide the orchestration abstractions of those broader frameworks (Rezazi et al., 19 Sep 2025, Raisanen et al., 2020).
Privacy is similarly partial by design. B5GRoam prevents on-chain disclosure of usage details, but other privacy surfaces remain salient in B5G roaming, such as location inference from gateway selection or mobility patterns (Tomasin et al., 2021, Tabassum et al., 2018). This suggests that a complete privacy-preserving roaming architecture would need to integrate settlement privacy, routing privacy, and mobility-aware control.
The supplied summaries also point toward future B5G environments in which roaming encompasses not only connectivity but computation, edge resources, AI/ML services, robotic control, and localized execution platforms (Raisanen et al., 2020, Liu et al., 2020). In that setting, B5GRoam can be read as part of a larger transition toward verifiable, policy-driven, and programmable inter-domain service exchange. The present literature most concretely establishes the settlement component: a zero-trust, on-chain, privacy-preserving, and rollup-scaled mechanism for reconciling roaming usage between operators without exposing sensitive CDR data (Rezazi et al., 19 Sep 2025).