1
From Open RAN to Open Spectrum: A Programmable, Intelligent Architecture for Multi-Service Spectrum Coexistence
arXiv:2609.11843v1 [cs.NI] 10 Sep 2026
Michele Polese, Senior Member, IEEE, Minh Dat Nguyen, Member, IEEE, Paolo Testolina, Member, IEEE, and Tommaso Melodia, Fellow, IEEE Abstract—Considering sharing or coexistence from the perspective of spectrum alone fails to recognize that any spectrum-enabled service also requires (i) radio and processing infrastructure and (ii) a protocol stack, including waveforms and signal processing pipelines. The efficiency of spectrum coexistence frameworks such as Citizen Broadband Radio Service (CBRS) is thus limited to optimizing resource allocation across a single dimension. How to address this limitation, however, remains an open challenge, especially considering the diversity of requirements and operational modes across spectrum services (e.g., sensing, communications, navigation, or positioning). This article introduces Open Spectrum, an architecture that brings softwarization, programmability, and open interfaces to heterogeneous spectrum services, extending the open Radio Access Network (RAN) principles beyond wireless networking. We propose to combine spectrum, services, and infrastructure in a common pool. Its resources are shared and orchestrated by a Spectrum Intelligent Controller (SIC), with plug-and-play spectrum applications, i.e., Spectrum Applications (sApps), and datadriven Radio-Frequency Interference (RFI) modeling using Digital Twins (DTs). We describe the Open Spectrum architecture, shared infrastructure pool, and operational workflows for tenant onboarding and incentives, conflict resolution, and service sharing across sensing, radionavigation, radiolocation, and cellular systems. System-level simulations using the BostonTwin urban DT and Sionna ray tracing show that there exist performance-driven incentives in sharing infrastructure and sharing across multiple services, enabling increased access to spectrum M. Polese, M. D. Nguyen, P. Testolina, and T. Melodia are with the Institute for Intelligent Networked Systems, Northeastern University, Boston, MA 02115, USA (e-mail: {m.polese, minhd.nguyen, p.testolina, t.melodia}@northeastern.edu). This work was partially supported by the U.S. NSF under award CNS-2434081, by the U.S. Government under Other Transaction number W15QKN-21-9-5599 between the National Spectrum Consortium (NSC) and the Government, and by OUSW(R&E) through Army Research Laboratory Cooperative Agreement Number W911NF-24-2-0065. The U.S. Government is authorized to reproduce and distribute reprints for Governmental purposes notwithstanding any copyright notation herein. The views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing the official policies or endorsements, either expressed or implied, of the U.S. Government. DISTRIBUTION STATEMENT A. APPROVED FOR PUBLIC RELEASE; DISTRIBUTION IS UNLIMITED
RF Sensing Telemetry
Radionavigation
NTN Sat
Sensing with cellular reference signals
High-density and bandwidth beacons
Radiolocation Cellular PHY-based positioning ISAC
Shared Infrastructure Pool
• •
Flexible basebands Programmable RF
Fig. 1: Use cases for spectrum coexistence with heterogeneous services based on a shared infrastructure, services, and spectrum pool.
and improvement in median Signal to Interference plus Noise Ratio (SINR) of up to 12 dB.
I. I NTRODUCTION Exclusive spectrum licensing, where a regulator assigns a frequency band to a single service or operator, who builds dedicated infrastructure to use it, is increasingly inadequate. The rise of Integrated Sensing and Communications (ISAC), high-precision positioning, space networking, and sensing creates demand for spectrum resources that static allocation cannot accommodate [1]. At the same time, unlicensed access does not provide guarantees to services with sensitivity to RadioFrequency Interference (RFI) (e.g., passive Earth exploration, radioastronomy, and sensing). Access requirements are also dynamic and shift over space and time: radioastronomy observations, for instance, benefit from bands outside established allocations due to Doppler shifts from moving targets. Regulatory and industry innovations have begun to address this. Frameworks such as Citizen Broadband Radio Service (CBRS), Dynamic Spectrum Access (DSA), and Automated Frequency Coordination (AFC) enable dynamic frequency sharing, albeit with a fundamental limitation: they address spectrum in isolation, without considering the infrastructure and services that depend on it. A CBRS Spectrum Access System (SAS) can grant frequency access, but it cannot help a radar operator who lacks the dense infrastructure to deploy a new sensing service, nor can it coordinate the joint use of radio hardware across service types.
This article introduces Open Spectrum, an architecture that extends the common-pool resource perspective on spectrum governance [2] to infrastructure and waveform sharing. In the common pool, resources are available to users of a heterogeneous set of services (tenants, as shown in Fig. 1). They are willing to share one or more elements among spectrum allocations, which can be based on legacy systems, and radio infrastructure; and to provide their service on a shared portion of the spectrum. This creates economic incentives for participation: sparse services (sensing, radar, navigation) gain access to dense infrastructure they could not economically deploy on their own, while telecom operators can dynamically extend their spectrum footprint for connectivity. The Open Spectrum system architecture provides observability as well as programmability through a closed control loop that (i) exposes spectrum and infrastructure telemetry; (ii) generates service-level policies beyond RFI alone; and (iii) enforces allocations on a programmable radio infrastructure. A shared infrastructure enables granular conflict resolution across heterogeneous services and, beyond spectrum, allows the same hardware and programmable basebands to deliver multiple spectrum services simultaneously (sharing services). In the remainder of this article, we compare Open Spectrum with prior literature and detail the proposed architecture. We then evaluate the opportunities associated with spectrum and infrastructure sharing through a Digital Twin (DT) framework that combines large-scale Ray Tracing (RT) with Sionna [3] with a real-world urban 3D model and multi-service deployments footprints [4]. Our results show the benefits of joint spectrum and infrastructure sharing. Services with cellular-like Radio Frequency (RF) parameters gain up to 12 dB in median Signal to Interference plus Noise Ratio (SINR). The orders-of-magnitude duty-cycle gap between sparse and cellular services leaves substantial idle time-frequency resources, which Open Spectrum reclaims through coordinated scheduling without degrading incumbent Quality of Service (QoS).
sharing. Current database-driven systems (CBRS [9], DSA [10]) make largely grant-or-deny spectrum decisions with limited QoS awareness, leaving sharing beyond spectrum unaddressed. Spectrum sharing without infrastructure or service sharing. Operators leverage Dynamic Spectrum Sharing (DSS) for flexible 4G/5G carrier allocation and neutral hosting [11] for shared cellular infrastructure. Surveys have catalogued sharing techniques across 5G bands [1], and spectrum governance has been studied through common-pool resource theory [2]. This identifies boundary rules, proportional allocation, and conflict resolution as prerequisites for sustainable sharing. Open Spectrum operationalizes these elements across heterogeneous services, unlike any existing spectrum architecture, which shares spectrum but not infrastructure or services. Service-specific resource models prevent crossservice optimization. Existing spectrum sharing frameworks define resources using service-specific abstractions: CBRS relies on Priority Access Licenses and General Authorized Access tiers [9], DSS operates over Long Term Evolution (LTE)/New Radio (NR) resource blocks [1], and satellite-terrestrial coordination relies on geometric exclusion zones [5]. These representations are fundamentally incompatible: they encode different notions of interference, time-frequency granularity, and protection criteria, preventing unified optimization across communication, sensing, navigation, and radiolocation under a common resource abstraction. III. O PEN S PECTRUM A RCHITECTURE The Open Spectrum architecture enables joint sharing of spectrum, services, and infrastructure through three interconnected components: the Spectrum Intelligent Controller (SIC), a shared infrastructure pool, and a spectrum monitoring framework. Figure 2 illustrates the overall architecture. Pool tenants request access to the SIC, specifying coexistence or frequency requirements, time, bandwidth, and other relevant parameters. The SIC processes such requests to provide access with guarantees, i.e., satisfying the coexistence or QoS requirements for the tenant. Spectrum Applications (sApps) plug into the SIC to implement modular spectrum policies. The shared infrastructure pool (bottom) provides the physical resources—compute, radio hardware, and sensors—that services access through the SIC’s coordination. Tenants can also submit best effort service requests to the infrastructure, without guarantees that, if the request is accepted, the service’s RFI requirements are satisfied. A DT module feeds site-specific propagation intelligence into the SIC’s decision engine.
II. R ELATED W ORK We identify three gaps in prior work that Open Spectrum addresses. Pairwise coexistence without multi-service coordination. Most spectrum sharing research examines coexistence between two system types: sensingcommunication pairs above 100 GHz [5], passive sensing interference [6], and O-RAN-based CBRS coexistence [7]. Multi-service works in the upper midband [8] address band-specific allocation but do not provide a unified cross-service controller or joint infrastructure 2
Requests Request 1 •coexistence requirements •time/location •bandwidth
Pool Tenants Telco 1
Deployment with guarantees m tru ec Sp
o inf
Spectrum Intelligent Controller Real-Time Spectrum Twinning sApp
Allocation and Deconflicting sApp
Measurements + Maps/Models
+
+
Automated Service Deployment, Configuration, Update
RF Sensors and Infrastructure-as-a-Sensor • • •
Other Tenants
… DoD Radar
Inter-SIC Coordination
Spectrum Compute, sites Requests Allocations
Shared Infrastructure Pool EESS Experiment
=
To other O-RAN systems (e.g., RICs)
Direct deployment on available infrastructure and spectrum (best effort)
Flexible basebands Programmable RF Extends O-RAN O-Cloud
Area 1
Spectrum Intelligent Controller
Area 2
Fig. 2: Open Spectrum architecture: the SIC extends the O-RAN RIC to coordinate heterogeneous services via sApps, with DT-based RFI modeling and a shared infrastructure pool.
isfy δX + δY ≤ 1 for every co-channel pair—a necessary condition for interference-free operation when the SIC schedules non-overlapping time slots with appropriate guard intervals. If violated, it invokes a priority table to determine which tenant’s duty cycle to reduce and issues updated grants. Conflicts between sApps are resolved by the SIC’s arbitration layer with a configurable precedence order. A key proposed capability is the integration of a DT for data-driven RFI modeling [3, 4]. Rather than relying on static exclusion zones as in current SAS implementations, the SIC would use a site-specific DT that pre-computes propagation maps and predicts SINR at affected receiver locations. This can enable QoS-aware allocation decisions that account for actual building geometry, material properties, and multi-path propagation, moving beyond conservative worst-case thresholds. The evaluation in Section V uses this DT framework for offline characterization; closing the real-time feedback loop between the DT and SIC is a subject of ongoing work.
A. Spectrum Intelligence Controller The SIC is the central coordination entity of Open Spectrum, enabling spectrum coordination for heterogeneous service types. It extends the closed-loop control introduced in O-RAN by the RAN Intelligent Controller (RIC) to control and data that have no representation in the O-RAN RIC, e.g., radar pulse schedules, navigation beacon timing, and sensing duty cycles. SICs operate on timescales of milliseconds (conflict detection) to hours (scheduled pool allocation). A SIC can also interface with O-RAN components, e.g., the Near-RT RIC, to enforce policies that require tuning or configuration of the cellular Radio Access Network (RAN) beyond spectrum, compute, and radio resource allocation, e.g., cellular user load balancing or handover. Multiple SICs coordinate across geographic areas in a peer-to-peer fashion, enabling regional spectrum policies (e.g., urban vs. coastal) while maintaining local autonomy under a common national framework. Each SIC maintains a local state database of active tenants, their current grants, and the DT’s propagation cache for its coverage area. When a sharing request involves a boundary region between two SICs, an exchange is coordinated across the two SICs, which evaluate shared priority of spectrum and infrastructure access for the stakeholders submitting conflicting requests. Compared to a hierarchical approach, this enables scalability and low latency for cross-boundary interactions. Spectrum Applications (sApps). The SIC supports plug-and-play customization through sApps—modular policy functions that third parties (regulators, service providers, researchers) can deploy and update independently. For example, a duty-cycle enforcement sApp ingests each tenant’s activity profile (δi , frequency band, geographic area) and verifies that duty-cycle budgets sat-
B. Shared Infrastructure Pool Open Spectrum extends the O-RAN O-Cloud concept into a shared infrastructure pool. The rationale is that sparse services (sensing, radar, navigation) lack the economic scale to build dense deployments, while cellular operators have (i) excess infrastructure capacity during off-peak hours and (ii) a need for additional spectrum during peak hours. Joint sharing creates value for both sides. The pool includes three resource categories: • Compute: Programmable servers hosting virtualized, software-based signal-processing functions, either for RAN, sensing, or other spectrum users. An automation layer [11] manages the service deployment, configuration, and life cycle. 3
Radio hardware: Radio units, software-defined radios, and reconfigurable hardware that can be dynamically reassigned, e.g., wideband Softwaredefined Radio (SDR) platforms at cellular sites configured for radar waveform generation during low-traffic periods. • RF sensors: Distributed spectrum monitoring devices for regulatory compliance and real-time conflict detection.
(duty-cycle adjustment), spatial separation (beam steering, power control), frequency reassignment, or prioritybased preemption. The SIC’s priority tables, configured per regulatory regime, encode service asymmetries: federal incumbents may receive preemptive protection, while commercial services negotiate shared access. This is enabled by the control that the SIC retains of the shared infrastructure pool: the architecture enables a direct sensing-to-enforcement loop. As a concrete example, consider a weather radar (sensing) and a 5G base station sharing the 3.5 GHz band in the same geographic area. If the radar begins a scan that would raise interference at the base station above the cellular tenant’s coexistence threshold, the SIC’s duty-cycle enforcement sApp first attempts temporal separation: it verifies whether the radar’s low duty cycle (δ ≈ 0.01) leaves sufficient residual capacity for the cellular tenant and, if so, schedules non-overlapping slots. If temporal separation is infeasible—e.g., because a third service already occupies the complementary slots— the SIC escalates to frequency reassignment, moving the cellular carrier to an adjacent channel. Throughout, the DT evaluates each candidate strategy before enforcement, reducing trial-and-error reconfiguration.
•
The pool operates in two modes. In scheduled mode, resources are reserved in advance for services with predictable requirements, e.g., radar pulse schedules with known dwell times and revisit intervals. The SIC precomputes a conflict-free time-frequency plan and distributes it to all affected tenants before the scheduling epoch begins. In best-effort mode, remaining capacity is dynamically allocated to services that tolerate variable access, such as Internet of Things (IoT) sensors performing periodic environmental measurements that can defer transmissions by seconds without impact. C. Spectrum Monitoring Open Spectrum leverages two complementary monitoring approaches: dedicated spectrum sensors providing ground-truth measurements of spectral occupancy analogous to the CBRS Environmental Sensing Capability (ESC) [12], and an “infrastructure-as-sensor” model where cellular base stations, radar receivers, and navigation beacons continuously report received signal characteristics to the SIC [13]. This combination provides both high-fidelity spectral measurements and broad spatial coverage, feeding into the DT’s interference predictions. The monitoring data serves a dual purpose: in the short term, it triggers conflict detection when observed interference exceeds tenant thresholds; in the long term, it calibrates the DT’s propagation models against ground truth, progressively improving the accuracy of predictive allocation decisions.
B. Service Sharing Open Spectrum enables deploying non-cellular services on existing cellular infrastructure, as illustrated in Fig. 1. Three modes are available: site sharing only (own RF equipment at cellular sites), full RF-chain sharing (adopting the host’s antenna, power, and height parameters), and spectrum pooling with duty-cycle coordination. The benefit depends on the match between native and host RF parameters: as we quantify in Section V, services with cellular-like parameters see the greatest gains, while high-power services may experience degraded performance under site-sharing only, due to increased interference (unless managed). The SIC selects among modes based on QoS requirements and the DT’s compatibility assessment. V. S YSTEM -L EVEL E VALUATION We validate the quantitative premises underlying the Open Spectrum architecture through system-level simulations based on the DT component of the SIC, with the goal of characterizing the benefits and tradeoffs of infrastructure and spectrum sharing in Sections III–IV. They do not constitute end-to-end validation of the SIC, sApps, or DT feedback loop, which requires a full-stack prototype and is the subject of ongoing work.
IV. O PEN S PECTRUM O PERATIONS The Open Spectrum architecture enables two key operations for multi-service, multi-dimensional sharing. A. Conflict Detection and Resolution The SIC continuously monitors the radio environment and compares observed conditions against each tenant’s coexistence criteria. When a conflict is detected—through predictive modeling (DT) or real-time monitoring—the SIC initiates resolution on timescales of tens to hundreds of milliseconds. Resolution strategies selected by the active sApps include temporal separation
A. Simulation Setup and Evaluation Scenarios The simulation framework integrates the BostonTwin urban DT [4] with the Sionna ray-tracing engine [3] for 4
TABLE I: Per-Service simulation parameters, based on spectrum regulations from FCC, NTIA, and ITU, and technical specifications from relevant standard-development organizations. Note that all services overlap by at least 400 MHz in the spectrum between 3.1 GHz and 3.7 GHz, which is used as reference for the common pool in this paper. Gain
Tx Pwr
Duty
Bandwidth
Typical Frequency Range
Nodes
Tx Height
Sensing Radionavigation Radiolocation Cellular
15–30 dBi 20–28 dBi 30–40 dBi 15–29 dBi
40–50 dBm 60–75 dBm 70–90 dBm 40–46 dBm
1–10% 0.1–1% 0.1–1% 70–100%
200–400 MHz 100–200 MHz 200–400 MHz 20–100 MHz
3.1–3.7 GHz 2.7–5.65 GHz 2.7–5.65 GHz 2–4 GHz
10/50 10/50 10/50 50/100
30 m 50 m 100 m 25 m
CDF
Service
1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 −30
Sensing – Baseline Sensing – Full-RF-chain sharing Sensing – Site-only sharing RadNav – Baseline RadNav – Full-RF-chain sharing RadNav – Site-only sharing Radiolo – Baseline Radiolo – Full-RF-chain sharing Radiolo – Site-only sharing Cellular
−20
−10 0 SINR [dB]
10
20
Fig. 4: CDF of downlink SINR: baseline (solid), full-RF-chain sharing (dashed), site-only sharing (dotted). Cellular in black.
transmitters treated as interference throughout. Table I summarizes the per-service parameters (sourced from U.S. and international spectrum policy regulations), while Fig. 3 shows a realization of the multi-service deployment. We consider a 400 MHz shared spectrum pool centered around 3.5 GHz and a thermal noise floor of −174 dBm/Hz. We evaluate three configurations. In Baseline, each service operates on its own exclusive band with dedicated infrastructure (current state of the art); specifically, services are centered at fc = 3.5 GHz (cellular), 3.4 GHz (sensing), 2.8 GHz (radionavigation), and 3.3 GHz (radiolocation), so cross-service interference is absent by construction. With Infrastructure sharing, a sparse service reuses cellular deployment at fc = 3.5 GHz, either adopting cellular RF parameters (full-RF-chain sharing) or deploying its own equipment at cellular sites (site-only sharing). We assume that, in the case of sharing, the SIC can coordinate access without generating harmful RFI across services.
Fig. 3: Multi-service deployment in BostonTwin. Triangles: base stations; dots: user equipment. Colors distinguish services.
site-specific evaluation. BostonTwin provides accurate 3D building geometry for a 2.3 km2 urban tile, loaded into Sionna RT with radio material properties (permittivity, conductivity) assigned per building. Antenna radiation patterns follow 3GPP TR 38.901 specifications. Results are averaged over 50 Monte Carlo iterations with uniformly sampled user locations. The deployment densities reflect the asymmetry between cellular and sparse services: cellular deploys 50 base stations in 1 km2 , while each sparse service uses 10 nodes in the baseline. This density asymmetry matters for the siteonly sharing case. Sparse services that keep their native, higher transmit power while operating from the denser cellular grid generate far more aggregate interference than from the baseline deployment. Therefore, in fullRF-chain mode, their power is set to cellular levels. The simulation captures inter-service interference by computing aggregate received power from all co-channel transmitters, including cross-service contributions (e.g., cellular base stations interfering with radar receivers and vice versa), providing a realistic assessment of coexistence tradeoffs. The SINR is modeled per service: cellular downlink follows the standard linear receiver; active sensing uses the two-way radar equation [14]; radionavigation and radiolocation report post-correlation SINR at the strongest-beacon receiver, with cross-service
B. SINR Performance under Infrastructure Sharing Figure 4 presents the CDFs of downlink SINR under three infrastructure configurations. The baseline configuration groups the three CDFs for sensing, radionavigation (RadNav), and radiolocation (Radiolo) around medians within −9 and −8 dB. Sensing shows higher SINR (i.e., the power over noise and interference at target) for farther away targets compared to radio services for navigation and location. Under site-only sharing, the higher transmit power for sensing, location, and navigation when compared to cellular leads to increased 5
Sensing Radionavigation Radiolocation
400
200
50
100
200 400 Bandwidth (MHz)
800
at δ ∼ 0.01 and B ∼ 210 MHz, and radiolocation (◦) at δ ∼ 0.01 and B ∼ 400 MHz. These operating points sit well below the cellular regime (δ ∼ 0.9–1.0), confirming that sparse services typically use less than 10%, and often less than 1%, of available time-frequency resources. When the SIC can schedule non-overlapping time slots (i.e., δX + δY ≤ 1 with synchronized slot boundaries and guard intervals), spectrum sharing is interference-free. When duty-cycle budgets are incompatible or services have inflexible timing requirements (e.g., radars with fixed dwell schedules), the SIC resorts to frequency-domain or spatial separation. With four or P more co-channel services, the generalized constraint i δi ≤ 1 makes scheduling combinatorial, requiring the SIC’s automated coordination. These rate curves also quantify the economic case for spectrum sharing. A sensing service operating at δ = 0.1 leaves 90% of time-frequency resources unused; a radionavigation service at δ = 0.01 leaves 99%. Under Open Spectrum, the SIC can allocate these idle slots to other tenants, increasing aggregate spectral efficiency without degrading existing services. The surface in Fig. 5 further shows diminishing returns at wider bandwidths for low-duty-cycle services, suggesting regulators could specify maximum bandwidth–duty-cycle products rather than fixed band allocations. VI. D ISCUSSION The evaluation results reveal several insights that inform both the Open Spectrum architecture and broader spectrum policy. Infrastructure sharing is not one-size-fits-all. The different nature of incentives and performance gains for heterogeneous services points to the SIC operating as a matchmaker, not a simple resource allocator. Each sharing request requires a compatibility assessment that considers the full parameter vector—power, gain, duty cycle, antenna pattern, location, and height—rather than frequency alone. This contrasts with current SAS designs, which primarily manage frequency-domain access. Temporal complementarity is the key enabler. The duty-cycle analysis shows that most non-cellular services use less than 1–10% of available time-frequency resources. This temporal sparsity is the primary resource that Open Spectrum exploits: by coordinating timedomain access across services, the SIC can achieve interference-free coexistence without the power reductions or geographic exclusion zones that characterize current sharing frameworks. Hardware compatibility constrains sharing depth. Full-RF-chain sharing assumes that sparse services can adopt cellular RF parameters, but radar waveforms require high Peak-to-Average Power Ratio (PAPR) and
Effective Rate (Mbps)
Duty Cycle
1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0
0
Fig. 5: Cellular effective rate vs. duty cycle and bandwidth. Markers: sensing (×), radionavigation (□), radiolocation (◦).
interference when the services are deployed at the much higher density presented by the cellular network. FullRF-chain sharing, by contrast, removes the interference penalty. The combination of increased density and finetuned RF parameters (including transmit power) lead to an alignment between the cellular SINR and the other services, with a median improvement of 12 dB. These results quantify the incentive that the sensing, radiolocation, and radionavigation spectrum services may have in joining the common pool: access to highdensity infrastructure can lead to improved SINR, as long as RF parameters are properly tuned. The DT enables such tuning before deployment: by simulating each candidate configuration against BostonTwin’s 3D geometry, the SIC can predict which mode will meet a tenant’s QoS target and avoid costly reconfiguration. It is worth noting the asymmetry between the uplink and downlink perspectives. The results presented here focus on downlink SINR at user/receiver locations, but the interference landscape differs for uplink-dominated services such as passive sensing, where interference is experienced at the base station or sensor rather than at a distributed user population. Passive receivers cannot adjust their transmit power, so their coexistence relies entirely on the SIC scheduling active transmitters around passive observation windows—well-suited to the dutycycle enforcement sApp. C. Effective Rate vs. Duty Cycle and Bandwidth Figure 5 characterizes the cellular effective rate, defined as Reff = δ · B · C , where C is the median spectral efficiency from the ray-traced SINR distribution, as a function of duty cycle δ and bandwidth B , when sparse services share the same infrastructure as cellular under SIC-coordinated non-overlapping scheduling. The duty-cycle ranges in Table I are sourced from NTIA radar characterizations [15] and spectrum regulations [8], and the red markers indicate the nominal operating points of each sparse service: sensing (×) at δ ∼ 0.1 and B ∼ 400 MHz, radionavigation (□) 6
amplifiers exceeding the ∼46 dBm ceiling of cellular radio units, while navigation beacons use narrowband waveforms that may conflict with the NR subcarrier spacing. Full-RF-chain sharing is thus currently viable only for sensing-like services, whereas high-power services should begin with site-only sharing and transition as reconfigurable SDR front-ends mature. Economic viability requires cellular and passive sensing QoS preservation. Cellular operators will only participate in Open Spectrum if hosting additional services does not trigger QoS violations or customer complaints. The tight interaction between SIC and shared infrastructure is a key enabler for both. DTs bridge the gap between conservative and aggressive sharing. Current exclusion zones are deliberately conservative, protecting incumbents with large geographic margins that leave spectrum unused. Sitespecific propagation modeling can tighten these margins by accounting for actual building geometry and terrain, but this requires validated models and continuous calibration—capabilities that the Open Spectrum monitoring framework is designed to provide. The transition from static exclusion zones to dynamic, model-driven sharing boundaries represents a gradual deployment path rather than a binary switch. VII. C ONCLUSIONS
[2] M. B. H. Weiss, P. Krishnamurthy, and M. M. Gomez, “How can polycentric governance of spectrum work?” in Proc. IEEE Int. Symp. Dyn. Spectr. Access Netw. (DySPAN), Mar. 2017. [3] J. Hoydis, F. A. Aoudia, S. Cammerer, M. Nimier-David, N. Binder, G. Marcus, and A. Keller, “Sionna RT: Differentiable ray tracing for radio propagation modeling,” in Proc. IEEE Globecom Workshops (GC Wkshps), Dec. 2023, pp. 317–321. [4] P. Testolina, M. Polese, P. Johari, and T. Melodia, “Boston Twin: The Boston digital twin for ray-tracing in 6G networks,” in Proc. 15th ACM Multimedia Syst. Conf., Bari, Italy, Apr. 2024. [5] M. Polese, X. Cantos-Roman, A. Singh, M. J. Marcus, T. J. Maccarone, T. Melodia, and J. M. Jornet, “Coexistence and spectrum sharing above 100 GHz,” Proc. IEEE, vol. 111, no. 8, pp. 928–954, Aug. 2023. [6] P. Testolina, M. Polese, J. M. Jornet, T. Melodia, and M. Zorzi, “Modeling interference for the coexistence of 6G networks and passive sensing systems,” IEEE Trans. Wireless Commun., vol. 23, no. 8, pp. 9220–9234, Aug. 2024. [7] S. Acharya, S. Li, N. Jiang, Y. Wu, Y. T. Hou, W. Lou, and W. Xie, “Mitra: An O-RAN based real-time solution for coexistence between general and priority users in CBRS,” in Proc. IEEE 20th Int. Conf. Mobile Ad Hoc Smart Syst., Toronto, Canada, Sep. 2023, pp. 295–303. [8] P. Testolina, M. Polese, and T. Melodia, “Sharing spectrum and services in the 7–24 GHz upper midband,” IEEE Commun. Mag., vol. 62, no. 8, pp. 170–177, Aug. 2024. [9] M. Ghosh, “Continuing innovations in the CBRS shared spectrum band,” IEEE Wireless Commun., vol. 31, no. 5, pp. 12–13, Oct. 2024. [10] NTIA, “Advanced dynamic spectrum sharing demonstration in the national spectrum strategy,” Jun. 2024, accessed on Jun. 2026. [Online]. Available: https://www.ntia.gov/issues/nation al-spectrum-strategy/advanced-dynamic-spectrum-sharing-dem onstration-in-the-national-spectrum-strategy [11] L. Bonati, M. Polese, S. D’Oro, S. Basagni, and T. Melodia, “NeutRAN: An open RAN neutral host architecture for zerotouch RAN and spectrum sharing,” IEEE Trans. Mobile Comput., vol. 23, no. 5, pp. 5786–5798, May 2024. [12] WInn Forum, “Requirements for commercial operation in the U.S. 3550–3700 MHz Citizens Broadband Radio Service band,” Wireless Innovation Forum, Spectrum Sharing Committee WG1, Tech. Rep. WINNF-TS-0112-V1.9.2, Mar. 2024. [13] A. Lacava, L. Bonati, N. Mohamadi, R. Gangula, F. Kaltenberger, P. Johari, S. D’Oro, F. Cuomo, M. Polese, and T. Melodia, “dApps: Enabling real-time AI-based Open RAN control,” Comput. Netw., vol. 269, p. 111342, Sep. 2025. [14] M. I. Skolnik, Introduction to Radar Systems, 3rd ed. New York, NY, USA: McGraw-Hill, 2001. [15] S. Jones, R. Hinkle, F. Sanders, and B. Ramsey, “Technical characteristics of radiolocation systems operating in the 3.1-3.7 GHz band and procedures for assessing emc with fixed earth station receivers,” NTIA, Dec. 1999. [Online]. Available: https://www.ntia.gov/files/ntia/publications/ntia99-361.pdf
This article introduced Open Spectrum, an architecture that goes beyond current DSA frameworks by jointly sharing spectrum, services, and infrastructure. The SIC with plug-and-play sApps, shared infrastructure pool, and DT-based coordination address key limitations of existing approaches: pairwise-only coexistence, binary grant/deny decisions, and spectrum-only sharing. System-level simulations show the impact of infrastructure sharing incentives and that duty-cycle complementarity enables interference-free coexistence under SIC coordination, with cellular QoS preserved. This article serves as the foundational introduction for the Open Spectrum concept. The building blocks—ORAN, DTs, neutral hosting—exist today; realizing the Open Spectrum vision requires their integration alongside new standardization and policy efforts. Future work includes prototyping the system and evaluating scalability, resource allocation, security, and techno-economic implications. On the regulatory side, future frameworks should adopt service-agnostic grant structures specifying interference constraints rather than technology-specific operating rules. R EFERENCES
Michele Polese is a Research Assistant Professor at Northeastern University. He received his Ph.D. from the University of Padova in 2020. Minh Dat Nguyen is a Postdoctoral Researcher at Northeastern University. He received his Ph.D. from the Institut National de la Recherche Scientifique (INRS) in 2023. Paolo Testolina is a Research Scientist at Northeastern University. He received his Ph.D. from the University of Padova in 2023. Tommaso Melodia is the William Lincoln Smith Chair Professor at Northeastern University and Director of the Institute for Intelligent Networked Systems.
[1] A. M. Voicu, L. Simić, and M. Petrova, “Survey of spectrum sharing for inter-technology coexistence,” IEEE Commun. Surveys Tuts., vol. 21, no. 2, pp. 1112–1144, 2nd Quart. 2019.
7