E-RADIO Architecture Overview

• E-RADIO Architecture is a set of advanced designs spanning radio access, embedded systems, radio-over-fiber, and spatial sensing, enabling flexible and energy-efficient wireless systems.
• It integrates dynamic unicast/multicast selection, green energy optimization, and cross-layer SDR techniques to improve network performance and scalability.
• The architecture employs precise algorithmic and mathematical models, supporting real-time adaptability and cost-effective large-scale deployments.

The E-RADIO architecture encompasses several advanced designs and frameworks across radio access networks, embedded systems, radio-over-fiber transmission, spatial sensing, and neural backbone models. Despite their varied domains, all E-RADIO systems share a focus on optimizing network flexibility, energy efficiency, functional partitioning, and real-time adaptability. The following sections present a comprehensive survey of the principal E-RADIO architectures as formalized in leading research, with exact algorithmic, architectural, and mathematical detail as reported in the original sources.

1. E-RADIO in 5G/6G Radio Access Networks: Dynamic Unicast/Multicast and Terrestrial Broadcast

The Enhanced RAN for Dynamic Unicast/Multicast and Terrestrial Broadcast (E-RADIO) architecture, proposed to support 5G-era terrestrial broadcast within Next Generation Radio Access Network (NG-RAN), augments the 3GPP Rel-15 Cloud-RAN split by introducing a new centralized multicast (CU-MC) function (Säily et al., 2020). E-RADIO enables flexible, demand-driven selection between per-user unicast (point-to-point, PTP) and point-to-multipoint (PTM) multicast/broadcast transmission, addressing both radio and protocol inefficiency in legacy eMBMS.

Logical Architecture

• Core entities: gNB-CU-CP (control-plane), gNB-CU-UP (user-plane), gNB-DU, RRH, with the new CU-MC for PTM coordination.
• Splits: F1-C/F1-U (PDCP/SDAP and RLC/MAC/PHY), E1 (CU-CP <-> CU-UP), F1-M (CU-MC <-> DU for multicast control).
• RAN Broadcast/Multicast Areas (RBMA): Subsets of NR cells grouped by SINR similarity, adjacency or synchronization capability:

$$\mathrm{RBMA}_k = \{ c \in \mathcal{C}\mid \mathit{config}_k(c) = 1 \}, \text{ and } |\mathrm{SINR}(c_i) - \mathrm{SINR}(c_j)| \le \Delta_{\mathrm{SFN}}$$

• Dynamic transmission selection: For UEs $U=\{u_1,\dots,u_N\}$, the network computes

$$U_{\mathrm{mul}} = \sum_{u\in U} R_{\mathrm{PTM}}(\gamma_u), \quad U_{\mathrm{uni}} = \sum_{u\in U} R_{\mathrm{PTP}}(\gamma_u)$$

and schedules PTM if $$U_{\mathrm{mul}} - U_{\mathrm{uni}} \ge \Delta_{\mathrm{th}}$$ with at least $N_{\min}$ UEs.

• Protocol Enhancements: Addition of F1-M for multicast control, M1-NG (optional) for multicast UP, and explicit flows for area/sync configuration.

Functional Extensions and Performance

• Support: Free-to-air/receive-only (no uplink), large-area SFN (up to ≈120 km via negative-µ numerologies), robust MCS selection.
• Resource Savings: Dynamic SC-PTM/MC-PTM vs unicast yields 50–70% PRB savings at cell edges; SFN gains scale with $O(\sqrt{N})$ akin to MU-MIMO diversity.
• Latency: Control-plane ≈15 ms, user-plane ≈2 ms + sub-ms sync; RRC_INACTIVE reduces wakeup to O(10 ms).
• Fronthaul Scaling: RAN-SYNC metadata <1 kb/s per DU; F1-M ≈100 kb/s per SFN service, outperforming LTE-eMBMS in OpEx and scaling.

E-RADIO thus unifies cloud-RAN flexibility, multi-service support, and seamless unicast/multicast switching under minimal protocol overhead, while enabling large-area terrestrial broadcast and robust control (Säily et al., 2020).

2. E-RADIO as Energy-Efficient, Functionally Split Radio-over-Ethernet (GROVE/E-RADIO)

The E-RADIO conception in “Green Radio OVer Ethernet” (GROVE) formalizes C-RAN functional split combined with distributed renewable energy and mesh-based fronthaul (Pamuklu et al., 2020). The E-RADIO radio-over-Ethernet system models:

Core Elements

• RRHs: Simple analog front ends, each assigned to a DU.
• DUs and CU: Contain digital processing modules (DPEs), with local or centralized execution of user-related functions (URFs).
• Ethernet Mesh (RoE): Interconnecting fabric; routing and bandwidth dynamically decided per split/traffic.
• Energy model: Each node (CU, DU) features local solar and grid supply, with optimal draw/sell decisions.

Mathematical Model

• Variables: Functional split assignment ($m_{idft}$), DPE activation ($a_{dt}$), path selection ($l_{rt(x,y)}$), green energy usage ($s_{rt}^y$).
• Objective: Minimize total expected on-grid OpEx:

$U=\{u_1,\dots,u_N\}$0

• Constraints: DPE CPU load, activation coupling, assignment completeness, per-link bandwidth, battery SoC, strict path selection.
• MILP linearization: Converts quadratic path $U=\{u_1,\dots,u_N\}$1 split constraints for scalability; solution via Gurobi solver.

Operational Results

• OpEx savings: Jointly optimizing splits, routing, and RES yields 15–30% cost reduction vs static benchmarks.
• Scalability: Up to 40-node meshes are tractable given hardware RAM limits.
• Insights: Adaptive centralization exploits mesh slack; battery-aware processing preserves local energy during grid price spikes.

This architecture demonstrates how E-RADIO paradigms can leverage dynamic functional splits, green energy, and packet-based fronthaul for high performance–cost efficiency in next-generation RANs (Pamuklu et al., 2020).

3. E-RADIO in Embedded Cross-Layer Optimized SDR Networks

E-RADIO is also presented as a fieldable, cross-layer optimized software-defined radio (SDR) platform integrating hardware and software co-design for multi-hop ad hoc wireless networks (Jagannath et al., 2022).

Hardware Architecture

• Base platform: Epiq Sidekiq-Z2 (Zynq-7000 SoC: FPGA + ARM Cortex-A9).
• RF front end: AD9364 RFIC, 1–3.5 W ext PA, GPS, DC/DC power system.
• FPGA: 802.11b PHY in hardware; real-time carrier recovery, symbol timing.
• CPU: Runs custom embedded Linux, kernel module, user-space daemons, L2-3 plus cross-layer beaconing.

Cross-layer Algorithm

• Routing: Per-packet utility

$U=\{u_1,\dots,u_N\}$2

where $U=\{u_1,\dots,u_N\}$3 is link energy efficiency, $U=\{u_1,\dots,u_N\}$4 is queue backlog relief, $U=\{u_1,\dots,u_N\}$5 are node–destination distances, $U=\{u_1,\dots,u_N\}$6 is neighbor residual energy.

• Distributed scheduling: Each node selects next-hop and PHY rate to maximize $U=\{u_1,\dots,u_N\}$7, with MAC employing CSMA/CA.

Field Experiment Results

• Reliability: >99% at 1–5.5 Mbps, ≥1 km.
• Goodput: >0.8 per-link at low rates; 6–10 node mesh yields capacity ~0.6–0.7, minimal control overhead.
• Adaptability: Real-time route selection responds to link congestion, power depletion, topology changes with sub-second responsiveness.

Trade-offs: FPGA PHY vs CPU MAC/routing sets throughput ceiling at 11 Mbps; single-radio and DSSS/CCK limit spectral efficiency and diversity; extensions proposed for multi-radio, OFDM, ML-based adaptation (Jagannath et al., 2022).

4. E-RADIO: Elastic Digital-Analog Radio-over-Fiber (EDA-RoF) Modulation/Demodulation

EDA-RoF (Elastic Digital-Analog Radio-over-Fiber), also termed E-RADIO, achieves a flexible transition between analog and digital RoF by multi-order, cascaded, low-resolution quantization with time-division interleaving (Zeng et al., 23 Dec 2025).

System Structure

• Tx chain: Wireless input → Tx-DSP → DAC → EA (Rapp model) → IQ MZM → fiber.
• Kernel: For order $U=\{u_1,\dots,u_N\}$8, sequentially separates signal into digital segments $U=\{u_1,\dots,u_N\}$9, one analog segment $$U_{\mathrm{mul}} = \sum_{u\in U} R_{\mathrm{PTM}}(\gamma_u), \quad U_{\mathrm{uni}} = \sum_{u\in U} R_{\mathrm{PTP}}(\gamma_u)$$0, and a residual $$U_{\mathrm{mul}} = \sum_{u\in U} R_{\mathrm{PTM}}(\gamma_u), \quad U_{\mathrm{uni}} = \sum_{u\in U} R_{\mathrm{PTP}}(\gamma_u)$$1; time-division-multiplexed for physical transmission.
• Rx chain: Coherent Rx → ADC → Rx-DSP → stage-wise de-multiplex, sum $$U_{\mathrm{mul}} = \sum_{u\in U} R_{\mathrm{PTM}}(\gamma_u), \quad U_{\mathrm{uni}} = \sum_{u\in U} R_{\mathrm{PTP}}(\gamma_u)$$2 to reconstruct $$U_{\mathrm{mul}} = \sum_{u\in U} R_{\mathrm{PTM}}(\gamma_u), \quad U_{\mathrm{uni}} = \sum_{u\in U} R_{\mathrm{PTP}}(\gamma_u)$$3, recursively restore $$U_{\mathrm{mul}} = \sum_{u\in U} R_{\mathrm{PTM}}(\gamma_u), \quad U_{\mathrm{uni}} = \sum_{u\in U} R_{\mathrm{PTP}}(\gamma_u)$$4.

Theoretical Foundations

• Spectral efficiency: For $$U_{\mathrm{mul}} = \sum_{u\in U} R_{\mathrm{PTM}}(\gamma_u), \quad U_{\mathrm{uni}} = \sum_{u\in U} R_{\mathrm{PTP}}(\gamma_u)$$5 stages,

$$U_{\mathrm{mul}} = \sum_{u\in U} R_{\mathrm{PTM}}(\gamma_u), \quad U_{\mathrm{uni}} = \sum_{u\in U} R_{\mathrm{PTP}}(\gamma_u)$$6

• SNR scaling: Empirically,

$$U_{\mathrm{mul}} = \sum_{u\in U} R_{\mathrm{PTM}}(\gamma_u), \quad U_{\mathrm{uni}} = \sum_{u\in U} R_{\mathrm{PTP}}(\gamma_u)$$7

where $$U_{\mathrm{mul}} = \sum_{u\in U} R_{\mathrm{PTM}}(\gamma_u), \quad U_{\mathrm{uni}} = \sum_{u\in U} R_{\mathrm{PTP}}(\gamma_u)$$8 is quantization-to-analog ratio; $$U_{\mathrm{mul}} = \sum_{u\in U} R_{\mathrm{PTM}}(\gamma_u), \quad U_{\mathrm{uni}} = \sum_{u\in U} R_{\mathrm{PTP}}(\gamma_u)$$9.

• Elasticity: $$U_{\mathrm{mul}} - U_{\mathrm{uni}} \ge \Delta_{\mathrm{th}}$$0 renders incremental improvements negligible.

Empirical Results

$$U_{\mathrm{mul}} - U_{\mathrm{uni}} \ge \Delta_{\mathrm{th}}$$1

• Applications: Enables elastic tradeoff between spectral efficiency and SNR, e.g., cloud-RAN fronthaul, mmWave, and satellite links.

E-RADIO in this variant offers an exact, tunable path across the analog–digital RoF continuum with explicit performance relations (Zeng et al., 23 Dec 2025).

5. E-RADIO for Wideband Spatial Sensing: Reconfigurable HSCD Architectures

An E-RADIO architecture comprised of sparse antenna arrays and sub-Nyquist sampling (SNS) integrated in Zynq SoC is designed to perform real-time spatial sensing under highly variable spectral conditions (Gupta et al., 2021).

System Description

• Block flow: SAA + SNS RF front end → FPGA (Preprocessing/SAP, EVD, V_n extraction, MUSIC spectrum) ↔ ARM Cortex-A9 (DMA control, DPR orchestration).
• Signal model: Multi-band mixing, autocorrelation and sparsification, EVD for noise subspace extraction, followed by MUSIC or ESPRIT for DoA estimation.
• Dynamic Partial Reconfiguration: On-the-fly SW/AP loads new V_n and MUSIC implementations when active source count $$U_{\mathrm{mul}} - U_{\mathrm{uni}} \ge \Delta_{\mathrm{th}}$$2 changes, saving resources and power.

Resource/Performance Metrics

• Pipeline latency (FPGA full, L=4, ULA): 66–75 µs.
• Resource savings: DPR-based design cuts BRAM/DSP/logic by ~9% over static "all-in-one".
• Accuracy: For SNR ≥ 20 dB and ≥200 RF samples, NDEE <0.03 (corresponds to <5.4° error).
• Efficiency: SNS gives 5× ADC rate reduction vs Nyquist, with minimal loss.

This E-RADIO platform offers low-latency and energy-scalable spatial awareness for 5G/6G RAN and spectrum-sharing networks (Gupta et al., 2021).

6. E-RADIO as a Hierarchical, Modular RRM Framework

The Envisioned RRM Architecture for Dynamic, Intelligent and Optimized-radio (E-RADIO) defines a three-layer stack for context-aware, hierarchical, and plugin-based radio resource management (Dryjański et al., 2021).

Architectural Layers

• Specialized Solution Modules: Plug-in spectrum allocation, interference management, load balancing and traffic steering modules.
• Abstraction Middleware: Unifies vendor/RAT specifics, offering standard data models (ResourceBlock, Cell, UEContext) and APIs.
• Coordination Layer: Global RRM optimization, conflict resolution, policy enforcement, plugin management.

Optimization Problem

$$U_{\mathrm{mul}} - U_{\mathrm{uni}} \ge \Delta_{\mathrm{th}}$$3

subject to RB and power constraints, fairness, and minimum URLLC rate.

Interaction and Extensibility

• Plugin API: Hot-pluggable modules with documented capabilities integrated via registration/discovery APIs.
• Interoperability: Aligns with 3GPP F1/Xn/RRC interfaces and O-RAN A1/E2 service models.

While the architecture itself is a conceptual reference, referenced METIS-II/SON modules demonstrated empirical gains of 20–30% throughput in hotspots, Jain’s index >0.9, and sub-ms reconfiguration in small-cell tests (Dryjański et al., 2021).

---

E-RADIO thus denotes a set of functionally advanced, mathematically rigorous architectures tailored for adaptive, scalable radio access, transmission, and control, validated across C-RANs, eSDR networks, RoF platforms, spatial sensing systems, and network management frameworks. Each instantiation leverages precise hardware–software partitioning, cross-layer optimization, or modular plugin architectures, with performance guarantees formalized via explicit models and empirical benchmarks.