TruRanX — 5G NR PHY, Native to Arm and x86

WHAT WE DO

Four things, done to commercial-deployment standard

PHY

5G NR PHY Software

Production-grade Layer 1 for gNB and UE, built natively for both Arm and x86 — not ported between them — with an open look-aside interface for FEC on GPU, FPGA, or ASIC.

RAN

Open RAN Integration

O-RAN 7.2a fronthaul, FAPI-over-DPDK L1–L2 interfaces, and full-stack integration with open-source L2/L3.

NTN

Air-to-Ground & NTN

Physical and protocol-layer expertise for moving-platform radio — aircraft today, satellites next.

TOOLS

Diagnostics & Validation

Purpose-built instruments, like our 5G PHY Rx Inspector, for real-time visibility into a live RAN stack.

PROOF, NOT PROMISES

What's already running

A complete L1 + L2 + L3 gNB stack, running end to end on Arm Ampere + NVIDIA L4, tested against a commercial UE.

Runs across a wide range of hardware price points — from a desktop gaming PC with GPU to industrial server-class platforms.

Our LDPC decoder is included in Arm's official RAN Acceleration Library (RAL).

WHY TRURANX

What that adds up to

COTS-first

No proprietary silicon, no vendor lock-in.

Commercially proven

The UE LTE physical layer and protocol stack we built for SmartSky Networks (now Apcela) is commercialized — flying in a live network today, not sitting in a demo.

Scales both ways

The same software stack runs on an embedded Arm platform, a desktop gaming PC, or a multi-sector production server.

Boutique and deep-tech

A small, agile team with specialist depth across PHY layer processing, L2, and L3 development, and end-to-end commercial deployment experience.

ABOUT US

From stalled to flying — now doing it again for private networks and space

"We took a stalled air-to-ground physical layer to commercial deployment in under a year. That discipline shapes everything we build."

TruRanX is a wireless technology company built around one discipline: physical layer engineering that survives contact with the real world, not just a lab bench.

Our formative project was building the LTE UE physical layer — and hardening the Layer 2/3 protocol stack — for an air-to-ground (ATG) network for SmartSky Networks. The link had to hold synchronization from a business jet at cruise, across cells spanning hundreds of miles, through constant handovers, in spectrum shared with uncoordinated Wi-Fi traffic. That network — later acquired and relaunched as Apcela ATG — is flying today.

That experience shaped how we've built every physical layer since: native to both Arm and x86 — with the Arm-native build carrying the power-efficiency discipline that satellite payloads demand — software-defined so the accelerator is a deployment decision rather than an architectural commitment.

Our mission: one carrier-grade 5G RAN software stack, native to Arm and x86, serving both terrestrial private networks and satellite programs — without trading performance for silicon lock-in.

TECHNOLOGY

One PHY. Two native architectures. Any accelerator.

Our 5G NR PHY isn't ported between architectures — it's built natively for both Arm and x86. The native Arm build is what makes it satellite-friendly, where every watt is a payload constraint. The native x86 build — including desktop AMD chips using AVX-512 — puts the same performance on the COTS hardware terrestrial operators already run.

CPU-FIRST

General-purpose cores, Arm or x86, handle the control-heavy, lower-density baseband tasks and lightweight L1 processing.

GPU-CENTRIC

LDPC forward error correction and bit-level processing — the most data-parallel work — offload to a GPU, currently NVIDIA L4.

OPEN INTERFACE

FEC sits behind a look-aside interface that also supports FPGA or ASIC — the accelerator is a per-deployment choice.

DPDK-BASED

Radio and L2/MAC communication runs over DPDK, avoiding memcpy overhead and kernel intervention.

Full-stack, not just PHY

Our Layer 1 is FAPI-compliant and connects to Layer 2/3 over DPDK. We've integrated it end to end with open-source L2/L3 stack, into a complete Open RAN gNodeB — tested against a commercial Qualcomm 5G NR UE on Band n78.

Hardware Configuration
Arm-Native — Satellite & Power-Constrained Deployments
Embedded Armv9 SoC + GPU	e.g., CIX P1 class — DSP-competitive price point (~<$700), built for edge and SWaP-constrained deployments
Arm Ampere / AGI Server + NVIDIA L4	Production-grade multi-cell setup on server-class hardware
Share Same GPU or Add Extra GPU	Unlocks dedicated AI-RAN inference acceleration capacity
x86-Native — Terrestrial COTS Deployments
Server-Class x86 + GPU	Production-grade multi-cell deployment on infrastructure operators already run
Desktop-Class AMD (AVX-512) + GPU	A lower-cost x86 entry point for private-network and lab deployments, paired with a server-class NVIDIA L4 GPU or a consumer RTX gaming GPU

AI-RAN ready

Because FEC acceleration runs on a GPU rather than consuming the entire compute budget, the same hardware that runs our PHY has headroom left for low-latency AI-RAN inference at the edge — without a separate accelerator.

FOCUS AREAS

Where the physical layer is hardest

Open RAN & Private Networks

Standards-compliant, disaggregated RAN software for operators and enterprises who want performance without hardware lock-in. The native x86 build, including desktop-class AMD with AVX-512, means the PHY runs on compute infrastructure many terrestrial operators already have on hand.

Non-Terrestrial Networks

Our physical layer was forged on a moving platform — large cells, high Doppler, continuous timing drift, relentless handover, all in uncoordinated spectrum. Those are the same problems, at greater magnitude, that define satellite NR. For regenerative payloads, our native Arm implementation targets the constraint that matters most in orbit: performance per watt.

AI-RAN

GPU headroom on our DU platforms opens a path to running AI-driven network optimization workloads at the edge, alongside the PHY itself.

MILESTONES

Recent progress

JUNE 2026

Published the case for regenerative NTN built on air-to-ground lessons

An engineering analysis connecting our commercial air-to-ground deployment to the design requirements of a satellite-borne gNB.

MAY 2026

Completed full L1 + L2 + L3 gNB stack integration

Our Arm/NVIDIA L4-accelerated physical layer paired with Layer 2/3, tested end to end against a commercial UE.

MARCH 2026

Demonstrated 5G NR PHY on a Pegatron O-RU

Arm Ampere + NVIDIA L4 COTS hardware, validated at 256-QAM downlink / 64-QAM uplink with industry-standard test equipment. Introduced the 5G PHY Rx Inspector.

LEADERSHIP

Built by the people who deployed it

Raman Odugoudar

CTO & FOUNDER

Leads physical layer and RAN architecture at TruRanX. His team took a stalled air-to-ground physical layer to commercial deployment with SmartSky Networks, and has since built the company's Arm- and x86-native, accelerator-agnostic 5G NR PHY from the ground up.

Raman holds a master's degree in Digital Communication Engineering from Christian-Albrechts-Universität zu Kiel, Germany, and brings 20 years of experience in wireless communication system development and business development. A hands-on engineer at his core, he has a deep understanding of RAN technologies.

A 5G physical layer built for every form factor — from the edge to orbit