Recently mobile location based services (LBS) are provided via a mobile terminal such as a personal digital assistant (PDA), a cellular phone and so on. The location method that searches where the mobile terminal of a user is accurately is a key factor for providing convenient and useful LBS. The most well known method associated with positioning system is GPS. The signal of GPS satellites can be always acquired outdoors and this system provides comparatively accurate location information. However, there are difficulties in applying GPS directly to indoor positioning because of the weakness of signal. And the positioning using the mobile communication signal between a base station and a cellular phone does not provide adequate accuracy due to some technical limitations of communication systems when applied to navigation. In addition, severe multipaths are present in an indoor environment. So a new wireless communication technology is required for indoor positioning to achieve a better and appropriate accuracy. Wireless local area network (LAN) has been installed in a number of indoor areas such as office, terminal, campus, and park with interests in mobile Internet. Therefore, it is expected that a wireless LAN signal would be easily acquired for indoor positioning. Various positioning measurements like time of arrival (TOA), time difference of arrival (TDOA), and received signal strength may be employed for indoor positioning. Employing TOA or TDOA measurements requires a time synchronization between a transmitter and a receiver. But, a time synchronization between a transmitter and a receiver is very difficult if not impossible in a wireless LAN. Therefore, it is more appropriate to use signal strength for indoor positioning with a wireless LAN.

One method for indoor positioning uses a location fingerprint that can be an average value of signal strength over several seconds at one location. It stores signal strength of the APs as measured at each sample point in a database beforehand. It then compares signal strength of a mobile user with stored signal strength in the database to find a sample point nearest to the user. However, this method needs a number of sample points and requires to build the database of received signal strength at each sample point and to update it since any modifications of indoor structures change the signal strength at sample point. Another method is based on a propagation model that describes how a radio wave loses signal strength as it travels through an environment. The amount of loss is dependent on the propagation environment. The signal strength is typically modelled by using a log distance path loss model with a path loss exponent. The value of the path loss exponent depends on surroundings and building type. The range from a mobile user to an AP can be calculated by the signal strength loss over space. The location of mobile user is determined by triangulation after an estimation of more than three ranges from the user to APs. However, this method has a limitation in that a received signal strength changes over time because of an obstruction and multipath. It requires huge number of sample points and stores each sample point’s signal strength to compute an estimated path loss exponent in the propagation model in advance. Then the location of the mobile user is determined by using the propagation model with the previous estimated path loss exponent and the received signal strength of the user. The location accuracy is poor when a present path loss exponent in this model is not the same with the estimated path loss exponent since any modifications of indoor structure or human movements change the received signal strength at sample points.

Additional reference point can be installed with APs in wireless LAN to determine a location of a mobile user regardless of any changes in an indoor environment. Reference points and a mobile terminal of a user is needed to resolve a difficulty of inconsistently received signal strength over time due to changes of an indoor environment. The main idea of the reference point is generic enough to be used for other wireless network technologies as well. A reference point is chosen appropriately in the middle of an interested room or indoor area to receive signals with enough strength from all APs. The least square method estimates the path loss exponent in this propagation model. The path loss exponent should be determined for each AP. The signal strength of the APs as measured at additional reference points is used to estimate path loss exponents. And the ranges from the user to all APs are calculated by the signal strength of a mobile user and estimated path loss exponents. Finally the location of the user is determined instantly by triangulation. Differently with conventional propagation method, this method does not assume a path loss exponent in a propagation model as constant because it estimates the path loss exponent by using received signal strength at reference points when mobile user request his location in an indoor environment.

This indoor positioning can be implemented in two ways. One is handset-based implementation and the other is network-based implementation. Handset-based positioning system consists of three parts, which are one mobile terminal of a user, several APs and several reference points. The mobile terminal stores a known location of each AP and a location of each reference point in advance. Each reference point monitors signal strengths received from all APs. A mobile terminal also monitors them similarly at the same time. The mobile terminal estimates the path loss exponent in a propagation model with the received signal strength of reference points. Next, it determines the location of the user with an estimated propagation model and received signal strength of the mobile terminal and already known APs’ location and reference points’ location. Network-based positioning system consists of four parts, which are one mobile terminal of a user, several APs, several reference points and one additional positioning server. Positioning server instead of the mobile terminal stores known APs’ location and known reference points’ location in advance. Each reference point monitors signal strengths received from all APs and transfer the received signal strength to positioning server. At the same time received signal strength of the mobile terminal is transferred to positioning server similarly. The positioning server estimates a path loss exponent in a propagation model with already known APs’ location and reference points’ location and signal strength received from reference points. Next, it determines the location of the mobile user with an estimated propagation model and signal strength received from the mobile terminal.

A test bed for a field experiment is established on the sixth floor of the Automation and Systems Research Institute building in Seoul National University. The layout of this testing area is depicted in Fig. 1. It has dimensions of thirty meters by seventeen meters with ten different rooms. Three APs are installed at the locations indicated with star marks and three reference points are chosen in the middle of two rooms and a hallway at x marks. The signal strength of reference points received from all APs is collected. Three 3COM wireless LAN access points are used for APs in IEEE 802.11b infrastructure. And four Orinoco wireless LAN silver network interface cards are used for three reference points and one mobile terminal of the user. Mobile handsets for reference points and a mobile user are assumed as laptop computers. An HP laptop computer is used for the user and three Samsung laptop computers are used for three reference points. Each AP acts as a wireless signal transmitter and the reference point. The mobile terminal acts as the wireless signal receiver using a laptop computer with Lucent Technology Orinoco wireless LAN card. This wireless LAN card can detect the signal strength received from APs. The mobile user is assumed to go around two large rooms and a hallway, each of which has a reference point. An experiment has twelve test points and eight signal strengths received from APs are collected at each test point for two seconds.

Fig. 2 shows the location accuracy using three reference points for indoor positioning in the field experiment. First, the three APs’ location is represented as star marks and the three reference points’ location as x marks in this figure. Next, the true position of the mobile user is represented as circle marks and the estimated position of the user as square marks. The mean of location error is 2.8m and the standard deviation of location error is 1.7m from all test points.

The meter-level accuracy for indoor positioning is obtained from these results. This method has the advantages that there is no need to construct the database of the received signal strength of a number of sample points in advance and update it periodically. The obstruction such as wall, human and spatial structure in the indoor environment need not be considered for positioning as well. In summary, the location of a mobile user can be determined by indoor positioning method using wireless LAN received signal strength.