Location-Based Service (LBS) has recently gained a considerable amount of attention from both academic and industrial sectors. LBS can be summarized as the electronic contents service that is somehow related to user location. Locating objects and people outdoors is becoming trivial by using technologies like Global Positioning System (GPS), but due to indoor channel characteristics, accurately estimating location indoors remains a difficult problem. There are actually various ways for determining and tracking position indoors, but to do so accurately remain very costly. The proposed system adopts the assist GPS (AGPS) idea to calculate the user position inside a building using the roof-antenna (known position) that has Line of Site ( LOS ) with satellites. The antenna receives navigation data and sends it to the user inside the building using wireless network. The distance between the user and roof antenna is assumed to be less than 300m. In this case, the time delay between roof antenna and user is approximately equal to 1 Clear/ Acquisition (C/A) code chip period that detects using autocorrelation with error in cm. Real data from GPS receiver is used in simulation using Matlab program version 7. The effect of Geometric Dilution Of Precision (GDOP) on user position calculation is discussed for different GPS time, atellites numbers and combination. The error in user position calculation is directly proportional to GDOP, which decreases when the number of satellites increases. GDOP also increases when the user clock bias error is increased, especially when the number of satellites is less than 5; yet, it is not affected by user position. Generally, the error in user position using this method is small and acceptable for most indoor applications. This data is used to track usef movements inside a building through a path with a constant velocity. The performance of system under AWGN and effect of bias due to multipath and NLOS is improved using Extended Kalman filter (EKF). The filter has been successfully applied in many navigation systems. The results with EKF are compared with Nonlinear "Least Square method (Nonlinear LS) at different processing times and different noise levels. EKF gives high improvement compared with Nonlinear LS. The overall system with EKF is simple to implement and generally more accurate compared with other indoor geolocation methods.