Skip to content

Time Synchronization

Time synchronization is a fundamental aspect of DaaS-IoT systems and is required to ensure coherent behavior, reliable measurements, and consistent event ordering across distributed nodes.

Due to heterogeneous hardware, variable network conditions, and decentralized operation, time alignment cannot be assumed and must be explicitly handled by the system.

Motivation

Accurate time synchronization enables:

  • Consistent ordering of events across different nodes
  • Reliable logging and diagnostics
  • Distributed measurements and benchmarking
  • Correct interpretation of network latency and delays
  • Improved stability of distributed coordination mechanisms

Without synchronization, temporal correlation between events becomes unreliable, especially in multi-hop or highly dynamic networks.

Synchronization Challenges

Distributed and IoT environments introduce several challenges to time alignment:

  • Variable latency and jitter
  • Packet loss and retransmissions
  • Clock drift across heterogeneous devices
  • Multi-hop communication paths
  • Dynamic topology changes

These factors require synchronization mechanisms that are resilient and adaptive.

Synchronization Mechanisms

DaaS-IoT implements internal synchronization mechanisms based on the exchange of timestamped control messages.

The system estimates clock offsets and delays by observing round-trip timing and continuously refines these estimates to compensate for drift and network variability.

Synchronization is designed to operate transparently and does not require application-level intervention.

Mandatory Synchronization Phase

Starting from version 0.21.03, time synchronization is a mandatory prerequisite for bidirectional communication between nodes.

When a node attempts to communicate with another node:

  • An initial synchronization phase is automatically triggered
  • The system ensures that a minimum synchronization state is reached
  • Data exchange proceeds only after synchronization has been established

This behavior guarantees temporal consistency across the network.

Integration with Discovery

Time synchronization is tightly coupled with the discovery process.

The enforced sequence is:

  1. Discovery and admission of a node
  2. Assignment of identifiers (DIN and SID)
  3. Initial synchronization phase
  4. Normal data exchange

Synchronization is therefore not optional and cannot be bypassed when discovery is enabled.

Master Node Selection

In version 0.21.03, the selection of the synchronization reference node (master) follows updated rules:

  • The node that receives the discovery packet during the creation of a new network becomes the synchronization master
  • Master selection occurs only when a network is initially formed
  • Nodes joining an existing network automatically synchronize with the current master

This approach ensures a single, stable temporal reference within the network.

Timestamp Handling Improvements

Version 0.21.03 introduces improvements to internal timestamp handling:

  • Message headers and payloads are processed independently
  • High-precision timestamps are captured as early as possible during packet reception
  • Timing information used for synchronization is decoupled from payload decoding

These changes improve synchronization accuracy and system robustness.

Stability and Recovery

The synchronization subsystem continuously monitors synchronization quality.

If instability or degradation is detected:

  • Corrective adjustments are applied automatically
  • The system may roll back to the last stable synchronization state if necessary

These mechanisms help preserve long-term temporal coherence in dynamic conditions.

Note

Time synchronization behavior in version 0.21.03 differs from previous releases and is mandatory for correct system operation.