Edge-Cloud Solutions in Challenging Settings

We are an experienced multi-disciplinary engineering team focused on research & development of next-generation edge computing solutions. We work on a distributed & elastic basis, coordinating activities via internet workspaces & hubs. This enables us to tap a global talent pool as needed.  Our style is modelled on the Lockheed Martin Skunkworks® approach to innovation

We are creating a rugged computing system styled as a set of interoperable building blocks that can be deployed in fixed & mobile settings in challenging environments outside of pristine data centres

Our vision is a block can operate in the field as a Linux-enabled multicore server that can play numerous software defined roles & can be configured as Software Defined Infrastructure. Roles include an IoT gateway, a Network Function Virtualised (NFV) appliance, an execution platform for pretrained machine learning models, a node in a Software Defined Network (SDN). Applications will typically be containerised. At a higher level the blocks can be assembled to implement a Kubernetes Cluster, a Private or a Hybrid Cloud

The basic unit supports wired & wireless multichannel Gbps communications using industry standard PHY/MAC protocols to enable central cloud back-haul & peer connectivity

Other units support downstream connectivity to accommodate the variety of communications requirements encountered in industrial automation & similar Use Cases

Supported network topologies will include bus, ring, star & mesh arrangements

Each node can be factory configured regarding the balance of computing, memory & input/output capabilities to provide complementary capabilities to support edge-clouds with myriad possibilities

The size weight and power constraints (SWAP-C) involved in providing a rugged Linux server class computing system at the very edge of the network require a System on Chip (SoC) design that maximises   computing performance whilst minimising Thermal Design Power (TDP)

This argues for adoption of the ARM RISC rather than the Intel x86 CISC computer architecture

Our first design using a bought-in 2-core ARMv7 32-bit computing module was successfully applied in a challenging mining Use Case but required highly tuned C coding to meet the minimal viable product requirements

This led us to develop a next generation system based on a contemporary 16-core ARMv8 64bit SoC

This has shown promise in laboratory test conditions with creditable performance running out-of-the box open-source software whilst meeting SWAP-C mission requirements

 For example, we have been able to demonstrate Gbps packet processing & routing performance using Data Plane Development Kit (DPDK) in a Docker container hosted on a standard Linux distribution in a passively cooled rugged enclosure

This has informed our plans for a production system that uses ARMv9 SoCs in the pipeline that provide the substantially increased performance and enhanced security features to meet our full mission objectives