One failure can isolate a region.
Central towers and fixed fibre routes create critical points that can disappear through power loss, damage, or distance.
The autonomous brain for self-healing networks.
A proposed software layer that helps decentralized devices find a new route when the network around them changes.
The problem
Central towers and fixed fibre routes create critical points that can disappear through power loss, damage, or distance.
Moving vehicles, metal structures, weather, and changing terrain can make yesterday’s best path unusable.
Tightly coupled network systems make it difficult to combine the radios and edge devices already available.
The Tessera idea
Tessera is conceived as a hardware-agnostic routing layer. It observes the network, decides where a packet should go next, and adapts that decision as conditions change.
Continuously searches for a better path as nodes move, weaken, or disappear.
Designed to make routing decisions close to the network on low-power hardware.
Explores hand-offs between Wi-Fi, Bluetooth, RF, and satellite links.
Aims to protect critical messages when bandwidth becomes constrained.
How it is structured
Three layers turn raw network conditions into routing decisions that an operator can see and, when necessary, override.
Collects signal quality, battery health, movement, compute availability, and packet loss from every node.
Uses reinforcement learning to evaluate topology and choose the next hop for each packet.
Gives operators a live topology, network health, priority controls, and manual overrides.
A network that can recover
The first research milestone is a simulation where an operator can remove a virtual node and watch traffic find a viable route around it.
Where it could matter
Solar-powered nodes pass connectivity from home to home until traffic reaches a fibre or satellite gateway.
Teams, vehicles, drones, and temporary radios form a network that can reorganise when infrastructure is lost.
Robots and sensors maintain routes through warehouses, mines, and other environments where interference constantly changes.
Development path
Phase 01
Train and test routing behaviour inside a virtual 100-node network, including moving and failed nodes.
Phase 02
Run the model across a small cluster of low-power physical devices and deliberately introduce interference.
Phase 03
Measure latency, packet delivery, energy use, and resilience with a real operating partner in the field.
The ambition
Tessera is MathBrooks research into network resilience: transforming complex routing intelligence into infrastructure that operators can understand and communities can depend on.
Research with us