Skip to content
38,400 NODES · LIVE
Platforms DOC-REV 2026.02 · REFERENCE BUILD

P-0001 / Platforms

The edge-to-fleet stack, documented in one reference.

Pervasive Edge Linux (PEL), a 14-family System-on-Module hardware catalog, and the Pervasive Device Cloud — pre-validated together so your team ships an embedded product in weeks instead of quarters.

  • Stack layersFirmware · Hardware · Cloud
  • MaintainerPervasive Systems, Inc.
  • Release cadenceQuarterly LTS · monthly rolling
  • Ships fromBoulder · Munich · Singapore
FLEET TOPOLOGY · LIVE 38,412 nodes
N-01Sensor
E-12Edge AI
G-04Gateway
F-01Fleet ctrl
C-01Device Cloud
Uptime
99.982%
Onboarding
< 15 min
Region
us-west-2

§ 01 — Stack index

Three pillars. One integrated edge-to-fleet platform.

Every Pervasive deployment is built from the same three reference pillars. Each ships pre-validated against the other two, so firmware, hardware, and cloud integration are solved problems — not your team's next quarter.

  1. 01

    Pervasive Edge Linux (PEL)

    A maintained Yocto-based distribution with pre-validated Board Support Packages across all 14 System-on-Module families. BSP bring-up drops from weeks to under 48 hours.

    Base
    Yocto Kirkstone
    BSP bring-up
    < 48 h
    SoM coverage
    14 families
    Read the firmware spec
  2. 02

    14-family SoM catalog

    Industrial temperature System-on-Modules ranging from sub-1 W sensor nodes to rugged edge AI gateways. Every SKU ships with a validated PEL image and a long-lifecycle supply guarantee.

    Families
    14 SoM SKUs
    Range
    −40 °C → +85 °C
    Lifecycle
    10–15 yr
    Compare SKUs
  3. 03

    Pervasive Device Cloud

    Fleet orchestration that runs on-prem, in your VPC, or in our free public tier. Handles secure onboarding, OTA, telemetry, and policy at 38,400+ node scale without lock-in.

    Deployments
    On-prem · VPC · SaaS
    Free tier
    100 devices
    OTA success
    99.97%
    See the cloud layer

§ 02 — Firmware pillar

Pervasive Edge Linux (PEL): one distro, fourteen SoMs, under 48 hours to first boot.

PEL is a maintained, security-hardened Yocto distribution with Board Support Packages pre-validated across every SoM we ship. Your firmware engineers stop maintaining BSP forks and start shipping product.

  • BSP

    Pre-validated Board Support Packages

    14 SoM families share a unified BSP interface, so kernel, bootloader, and driver integration are validated before your team touches the bench. First-boot on a new SKU lands in under 48 hours, not the industry-typical three-week BSP cycle.

  • LINEAGE

    Yocto Kirkstone base, quarterly LTS

    Built on the current LTS Yocto release with a quarterly Long-Term-Support cadence and a monthly rolling channel for fast-moving teams. CVE patching is centralized — no per-SKU patch debt.

  • SEC

    Secure boot, measured root, signed OTA

    Hardware root-of-trust, signed and verified boot chain, dm-verity read-only rootfs, and signed delta OTAs are default — not a paid add-on. Used on three U.S. Department of Defense programs.

  • RTOS

    Real-time and container co-existence

    PREEMPT_RT kernel available for deterministic control loops, alongside a container runtime for application workloads. Run a real-time motion controller and a Python inference service on the same SoC without context-switch penalties.

§ 03 — Hardware pillar

14 System-on-Module families. Browse the four most-evaluated SKUs.

From sub-1 W sensor nodes to rugged edge AI gateways, every SoM ships with a validated PEL image, a 10–15 year lifecycle commitment, and an industrial temperature rating. The table below highlights the four SKUs most often evaluated in architecture reviews.

SKU CPU class RAM Storage Key I/O NPU / accel Temp Typical power
PEL-S1Sensor node Cortex-M33 @ 200 MHz 512 KB SRAM 8 MB flash I²C · SPI · UART · CAN-FD −40 → +85 °C 0.4 W
PEL-A4Industrial gateway Quad Cortex-A55 @ 1.8 GHz 2 GB LPDDR4 16 GB eMMC 2× GbE · USB 3.0 · RS-485 · CAN-FD −40 → +85 °C 4.8 W
PEL-X7Edge AI gateway Octa Cortex-A78AE @ 2.2 GHz 8 GB LPDDR5 64 GB eMMC 2× GbE · 4× USB 3.0 · 2× MIPI-CSI · PCIe Gen3 10 TOPS NPU −40 → +85 °C 11.5 W
PEL-R9Rugged AI node 12-core Cortex-A78AE + GPU 16 GB LPDDR5 128 GB eMMC · NVMe 4× GbE (TSN) · 2× 10 GbE · USB 3.2 · 2× CAN-FD 32 TOPS NPU + 1.6 TFLOPS GPU −40 → +85 °C 22.0 W

Full 14-SKU catalog (electrical specs, carrier-board reference designs, and lifecycle status) is reviewed during the Edge Architecture Review. Request the full datasheet

§ 04 — Cloud pillar

The cloud layer is orchestration, not lock-in.

Pervasive Device Cloud is the fleet-management plane that sits above PEL and the SoM catalog. It handles secure device onboarding, signed OTA campaigns, telemetry routing, policy enforcement, and alerting across every node in the field. It does not require your data to leave your perimeter to work.

The control plane runs in three deployment modes: a fully on-prem install for air-gapped facilities and regulated workloads; an in-VPC bring-up on AWS, Azure, or GCP for customers who want the orchestration layer inside their own cloud account; and a managed SaaS option for teams that want zero infrastructure to operate. The agent on every SoM speaks the same protocol regardless of where the control plane lives — switching modes is a configuration change, not a port.

For evaluation and prototyping, the free Pervasive Device Cloud public tier supports up to 100 devices indefinitely, with the same APIs and OTA semantics as the commercial tier. Over 4,300 developers are using it as a sandbox today; most convert to a paid commercial deployment once their fleet plan crosses the 100-device threshold.

Public Free 100 devices · indefinite · developer sandboxes
Managed Per-node SaaS control plane · 99.95% SLA
In-VPC Annual Helm chart into your AWS / Azure / GCP account
On-prem Annual Air-gapped install · FIPS-validated build path

§ 05 — Proven at fleet scale

Four numbers that prove the platforms run in production.

  • 38,400+ Edge nodes under active management As of February 2026
  • 99.982% Fleet uptime, trailing 12 months Across managed deployments, Dec 2025
  • 14 wk Average time-to-first-production vs. 65-week in-house benchmark (n=212)
  • 27 Countries with active deployments Four continents, three offices

Evaluate a Pervasive reference platform against your workload in a 30-minute architecture review.