Original article was published by RedNodeLabs on Artificial Intelligence on Medium
Wireless-Mesh Time Synchronization
A common clock source is required in a wide variety of connected things.
For instance, accurately timestamping sensor data coming from radars, LiDARs or accelerometers, is required for labeling datasets used for training in machine learning and sensor fusion applications. Similarly, in industrial applications, dependable and synchronized reaction must be assured in safety-critical wireless sensor and actuator networks. Tight synchronization is also the foundation of many real-time locating systems and time-division multiple access wireless communication schemes. A perfect schedule of data transmissions also unlocks an optimal wake-up scheme, in which battery life is maximized.
For computers working over unreliable networks, like the Internet, Network Time Protocol (NTP) is regarded as the state of the art, and can generally achieve synchronization within one millisecond. Advanced systems, like those used in particle accelerators, can achieve sub-nanosecond accuracy by using advanced dedicated hardware.
For low-power connected IoT devices, an NTP-capable network interface car is normally not available. Those that can afford being equipped with a GPS and operate outdoors can achieve accuracy within the microseconds range, depending on signal reception and synchronization frequency.
Simple IoT devices are potentially deployed indoors, have a minimal energy budget and cost must be kept to a minimum level. Typical synchronization schemes, based on low-power radios like Bluetooth or Zigbee rely on a node, acting as a master clock source, periodically sending synchronization beacons in a star topology.
Large networks working in harsh environments require multihop packet retransmissions in a mesh topology, and cannot trust a master synchronization node, which constitutes a single point of failure. Mesh networks are considered to be complex and unpredictable, complicating the synchronization task.
To fill this gap left by applications that require dependable wireless-mesh time synchronization, we created the most reliable flooding-based mesh network on the market: RedNodeBus.
RedNodeBus is able to achieve sub-microsecond network-wide synchronization accuracy over wireless mesh topologies, eliminating any single point of failure by decentralizing and distributing the synchronization task. Best of all: you do not need any dedicated hardware, we only use low-power Bluetooth radios, combined with advanced packet preamble detection and best-in-class synchronized flooding.
Focus on the core of your app, leave the time synchronization to us. We’ll take it off of your to-do list.