Hello,
I am currently working on a BLE Mesh–based control system and would appreciate guidance on a few architectural and scalability concerns.
System Overview
The network uses a star topology, with a central node (nRF52840 DK) interfacing with a PC via a serial connection. This central node receives commands from the PC and translates them into corresponding BLE Mesh messages (e.g., setting a level server or querying a sensor value from a specific node).
From a physical topology perspective, nodes are deployed linearly, spaced 40 meters apart, with the central node located anywhere in the network.
Provisioning Approach
Currently, provisioning and configuration are handled entirely by the central node. Upon receiving a serial command, the central node provisions and configures a device with the specified UUID within the mesh network. Node removal is handled through a similar process. The central node, therefore, maintains the network's Configuration Database.
This approach works well for small deployments (on the order of tens of nodes). However, we have concerns regarding scalability and robustness as the network grows.
Questions and Concerns
-
Provisioner State Backup and Recovery
Since the central node acts as the sole provisioner and stores the network’s CDB, is there a recommended way to back up or preserve this information?
Ideally, if the central node fails or becomes corrupted, we would like to restore the provisioner state to a replacement device and allow the network to continue operating without requiring re-provisioning of all nodes.
Would reading and restoring the entire flash contents of the central node be a viable or supported approach, or is there a more appropriate mechanism within the BLE Mesh stack for this purpose? Is this possible? -
Remote Provisioning
How would remote provisioning apply to a network with this topology and operational model?
Are there known limitations or best practices when using remote provisioning in large, linearly distributed networks? -
Provisioning via nRF Mesh App
Would migrating the provisioning process to the nRF Mesh mobile application be beneficial in terms of scalability, maintainability, or robustness compared to using a custom central-node-based provisioner?
Future Scale
While this product is still in development, the target deployment is to support hundreds of independent networks, each with tens of nodes, for a total of potentially thousands of devices.
Any recommendations or references to best practices for provisioning, backup strategies, and large-scale BLE Mesh deployments would be greatly appreciated.
Best regards, Santiago