Skip to main content

11. System Process

The system process contains mitigation algorithms that process sample data from the clock filter algorithm. These algorithms include selection, cluster, combine, and clock discipline algorithms.

11.1. System Process Variables

Key system variables include update time (t), system peer identifier (p), leap indicator, stratum, precision (rho), combined offset (THETA), combined jitter (PSI), root delay (DELTA), root dispersion (EPSILON), survivor list (v), reference ID, and reference time.

Key constants: NMIN = 3 (minimum survivors), CMIN = 1 (minimum candidates).

11.2. System Process Operations

11.2.1. Selection Algorithm

Identifies and eliminates falsetickers, preserving truechimers based on agreement principles. Produces a candidate list using interval intersection methods.

11.2.2. Cluster Algorithm

Refines the candidate list by removing statistical outliers. Iteratively discards the candidate furthest from the centroid until the minimum number of survivors remain. Selects one survivor as the system peer based on synchronization distance, stratum, and jitter.

11.2.3. Combine Algorithm

Computes final system offset and jitter using weighted averaging:

THETA = Σ(w_i * theta_i) / Σ(w_i)
where w_i = 1 / lambda_i

11.3. Workflow

  1. Input: Samples from clock filter algorithm
  2. Selection: Eliminate falsetickers
  3. Cluster: Refine via statistical methods
  4. Combine: Calculate weighted average
  5. Output: Final offset to clock discipline algorithm

Implementation example in Appendix A.5.5.6 clock_select() routine.