Toward Collinearity-Avoidable Localization for Wireless Sensor Network

Algorithm 3

LE-PCR.

Input: D: distance matrix between nodes. : coordinate of beacon nodes.

Output: : estimated location of the non-beacons.

  (1) Conduct standardization treatment to matrix .

  (2) To matrix after standardization treatment, use PCA to extract PCs and the score vector. (see Formula (15)).

  (3) Use conditional index to determine whether there is multicollinearity problem. If there is, eliminate corresponding

    PCs with a small characteristic root in accordance with the cumulative variance contribution rate.

  (4) Use the left PCs and PCR, and obtain the final location estimation through Formula (18).