Locating a Mobile station (MS) in wireless network iscurrently an important research topic. Instead of linearizing thenonlinear hyperbolic equations, a novel noise subspace approach isproposed for hyperbolic mobile location using the Multidimensionalscaling (MDS) analysis whose loss function is related to a scalarproduct matrix. This approach is further improved using the Weightedleast squares (WLS) algorithm. It is robust to large measurement noisedue to the fact that the dimension knowledge and eigen-structure of thescalar matrix are exploited. Computer simulations are included toverify the theoretical development and to contrast the estimatorperformance with the Two-step WLS (TWLS) method and the Cramer-Raolower bound (CRLB).