In order to achieve high accurate indoor positioning,this paper desig ns a visible light communication (VLC) based indoor positioning system.By combi ning the optimized Lambertian model,code division multiple access (CDMA) techno logy and trilateral positioning algorithm,the positioning accuracy and system s calability are effectively improved.To begin with,the ID information of each l ight emitting diode (LED) with the direct spread spectrum modulation is loaded t o the LED driver,and the ID bearing LED light signal is launched.At the receiv er,the optical signal is detected by the photodetector (PD) and the ID informat ion with the corresponding received signal strength (RSS) are recovered based on the orthogonality of the spread spectrum code that increases the channel capaci ty and enhances the anti-interference ability of the system.Then,using the Lam bertian model,the location estimation coordinates of the point to be located ar e calculated by the traditional trilateral positioning algorithm.To further improve the positioning accuracy,an optimal approach based on the k-nearest neighb or (KNN) is introduced.The RSS fingerprints of some appropriate points are coll ected and the Lambertian model parameters of each LED are renewed by these finge rprints.Thus,the optimized Lambertian model can be applied to obtain the highe r accurate positioning coordinates.The experimental investigation of the design ed VLC indoor positioning system was carried out in a space region of 1m×1m×1.35m.The results show that the average positioning error of the proposed high ac curate VLC indoor positioning system is reduced to around 2cm,and the positioni ng accuracy is improved by 30% as compared to the traditional trilateral positio ning algorithm.The proposed scheme based the Lambertian model optimization usin g RSS fingerprints has excellent scalability and can be applied in broad applica tions.