1. IntroductionIn this section we explain and describe Indoor Positioning Systems, as well as their basic classification and evaluation.1.1. What are Indoor Positioning Systems?An Indoor Positioning System (IPS) is used exclusively in indoor environments. Dempsey 1 defines an IPS as a system that can determine the position of someone or something in a physical space continuously and in real-time. From this we can conclude that IPS should always, unless it is turned off by the user.Location information provided by an IPS for location-based applications can be diverse, for example an absolute location information can be provided. In order for this to be possible, the map of the locating surface and area must be saved and available in the IPS. With regard to the map, the target’s absolute position can be displayed and measured. This type of positioning is often offered by the indoor navigation systems and indoor positioning tracking systems, since certain services, such as guiding and tracking services, need the exact positions of the targets in order to be accurate 2.On the other hand, the relative position information, which measures the motion of target’s different parts, is also the possible output of the IPSs. For instance, let’s suppose that we use an IPS that records whether the door of an oven is closed or not. This kind of positioning information is called relative position information, since recorded point on the door is positioned with respect to the oven itself 2.Proximity location information is a third type of position information, which determines location of the place where a target is situated. This kind of positioning is helpful in the situations where we do not need absolute, nor relative position information. For example, hospital systems for tracking and monitoring patients provide such information, particularly the room where a patient is, auditing whether the patient is in the wrong room or not 220.127.116.11. Classification of Indoor Positioning SystemThere are various ways to classify the IPSs, depending on the criteria. One idea is to classify them by the fact whether the certain IPS, in order to measure the position of an object, uses an existing wireless network infrastructure. In regard to this, IPSs can be classified as network-based approach and non-network-based approach. An advantage of the network-based approach is that there is no 1 need for the additional hardware infrastructure, since it uses resources from existing infrastructures. This type of approach is preferred, mainly for cost reasons. However, the non-network-based approach may offer higher accuracy, since its infrastructure for positioning is more flexible because it is designed to meet expectations and needs of this very approach 2.Another idea is to classify them according to their system architecture. There are three different architectures: infrastructure positioning architecture, self-positioning architecture and self-oriented- infrastructure-assisted architecture. The infrastructure positioning architecture measures targets’ positions with the help of the infrastructures, which can record and track the position of devices, under the condition that they are in the coverage positioning area. Next, self-positioning measures the positions of the targets by the targets themselves, which means that this type of positioning takes advantage of previously mentioned infrastructure positioning systems, which are equipped with privacy and high security. Finally, in the self-oriented and infrastructure-assisted architecture, in order to start position measurements and to get location information of a target, it is necessary for the tracked target to send a request to the positioning system 2.1.2. Criteria of Evaluating Indoor Positioning SystemsIn this section, various deployment criteria are described and all of them are fully focused on user experience and user preference. These criteria are used to evaluate the IPSs.1.2.1. Security and PrivacySecurity and privacy are crucial issues concerning users and their needs. This is true, since users care if someone records and tracks them and their activities, which is why they want to fully control usability of their location information and history. Privacy in IPSs can be improved by controlling the access and distribution of the location information, both from the system architecture side and software side. For example, self-localized position system architecture is a good choice regarding privacy, since it ensures it by performing location estimations in the target device and because of this the access is granted only if the target device gives to an entity its location information 18.104.22.168. CostThere are several parts of the IPS, which should be taken into account when evaluating its cost, such as the cost of a positioning device for each user, the cost of the infrastructure components and the cost of system installation and maintenance. For example, one of the expensive and complex positioning systems are GPSs, due to the fact that they have a large infrastructure to support the location measurement. On the other hand, some existing infrastructures, such as WLAN, can be reused by some IPSs, which is more cost-effective, since there is no need for new infrastructures and, thus, no need for extra costs 2.The common mistake users make is that, often, they specify the cost of the device when they buy the new device. However, they do not consider maintenance cost of the device, such as the life time and battery costs — for example, if a device has a longer battery life, it usually means that it has lower maintenance cost because it needs less frequency of changing the batteries. As mentioned before, devices with self-localized position are preferred because they offer higher security for the end users. 2 Because of this feature, the cost of devices like these is higher, indirectly because their battery life time duration is decreased, and directly because the device performs much more complex calculations for positioning 2.For the long-term use of the positioning systems, the cost of its installation and maintenance needs to be addressed. This can be done in different ways, for example space and time are important factors. Space cost considers place and size of where the user devices and installed infrastructure components occupy — either way, physically large positioning devices are not suitable for a user to carry it every day. On the other hand, if the system fails to work due to the serious faults, the time cost involves the time length of the positioning system and time requirements of system installation 22.214.171.124. PerformanceTwo main parameters for performance in IPSs are precision and accuracy. Accuracy stands for the average error distance, while the precision is the success probability of position estimations regarding the predefined accuracy. In addition, there is another performance aspect, which is the delay of IPS, or more precisely, “the delay of measuring, calculating positions of estimated target and forwarding position information to the requesting parts” 2. This is because the target that is being tracked might move quickly, but also because indoor environments are changing dynamically. Another issue regarding the performance is scalability, which is “defined as the number of objects that an IPSs can locate with a certain amount of infrastructure devices within a given time period” 2. Finally, there is a frequent trade-off between the performance and the price of an IPS — systems with higher performance have higher costs. For instance, adding filters to the positioning system reduces the sunlight influence, thus increasing the price of that system because of these extra filters. 21.2.4. Robustness and Fault ToleranceA robustness of an IPS is referring to the ability of the system to keep on operating, despite some serious problems, such as running out of battery energy. For example, it may happen that the sensors of the sensor-based positioning system in a public area are stolen. Even in this case, the system should still provide positioning information, although it might have a lower accuracy. 21.2.5. ComplexityEvaluation of complexity of IPSs involves the human interaction with the system and their intervention during the maintenance and deployment of the IPS. This means that systems with a rapid set-up and a small number of infrastructure components are preferred because of their simplicity. Computation time of the determination of user’s position is another aspect of the IPS complexity. Lower calculation time is preferred, often because of the low battery power and limited CPU processing of the mobile devices. 21.2.6. User PreferenceIPSs should meet users’ requirements, since they are designed and developed for them. The following specifications are usually desired by the users — “the devices should be wireless, small, light weight, lower power consumption and computational powerful to offer rapid, accurate and real- 3 time positioning services” 2. Finally, software and infrastructure components should be user interface friendly and easily learnt for the users. 21.2.7. Commercial AvailabilityAlthough there are research-oriented positioning systems, which are not available for commercial use, there are many existing commercially available IPSs. Designers of such systems are addressing numerous aspects in order to make them popular in the market. However, since there is a competition between companies in the market, many of them are keeping their commercial positioning products’ principles secret, which is not the case with the research-oriented systems, since their design is open for all 126.96.36.199. LimitationsDespite numerous improvements in the IPS technology, there are still some limitations. One of the most significant is the medium used in positioning. For example, as mentioned above, positioning systems that use WLAN technology are preferred, since their cost of positioning is reduced because of the reuse of the existing WLAN’s infrastructure. However, the increase of the error range may happen in the radio frequency based positioning, because of the multi-path and reflection effects. Another common limitation is the scope of the positioning system, since some of them are designed for covering a short range and, thus, they are not scalable for larger areas. Finally, some of the IPSs can not track large number of targets, since they are designed to be used by small number of people simultaneously 2.2. Location Technologies, Techniques and AlgorithmsThe purpose of IPSs is the location-awareness in computing systems. With increased need for them, a numerous wireless technologies have been developed for indoor locating, such as magnetic technology, IR, UWB, Bluetooth, ultra-sound, WLAN, RFID, etc. Each of these technologies has different advantages in indoor locating. Several IPSs wireless technologies will be discussed in the following sections 2.2.1. Indoor Position EstimationsAs mentioned in the introduction, relative, absolute and proximity location information are offered by IPSs, which can be equipped with one or several combined technologies. Currently, four techniques exist for indoor position estimations — triangulation, fingerprinting, proximity and vision analysis. “Triangulation, fingerprinting and vision analysis positioning techniques can provide absolute, relative and proximity position information. The proximity positioning technique can only offer proximity position information” 2. IPSs usually use one positioning technique, although there are IPSs which are designed as combination of two or more positioning techniques, in order to compensate limitations of single positioning technique 188.8.131.52. TriangulationFig. 1. Triangulation Positioning Techniques 2Triangulation method (Fig. 1) uses three different methods to calculate the position, based on the geometric properties of triangles — angle of arrival (AOA), time of arrival (TOA) and received signal strength (RSS) 3. The absolute position E1 can be calculated if the geographical coordinates (xi, yi) of A, B, C are known, by using either the directions or the length of R1, R2 and R3. All three types of position information can be calculated and derived using this method. Each triangulation method has advantages. TOA is the most accurate technique for indoor positioning, since it can filter out multi- path effects, but it is complex to implement. However, there are also limitations for triangulation method. For example, in order to estimate position of an object, TOA and RSS must know the positions of three reference elements, as seen in Fig. 1, A, B and C. Meanwhile, even though AOA needs only two position measuring elements, it might contain some errors if the object is located far away, resulting in lower accuracy. 4 22.1.2. FingerprintingFingerprinting positioning technique requires pre-measured location related data in order to successfully improve the indoor position accuracy. It consists of two phases: offline training phase and online position determination phase 5. Location related data is collected and measured for the 5 position estimation in the offline phase, while in the online position determination phase, location data of a target objects is measured and compared with the same data collected in the offline phase, in order to obtain a similar case in the database. For example, in an IPS 6, position estimation is calculated via WLAN technology. In Figure 2 (a), we can see that three access points (APs) are fixed in an area of 25 m x 25 m. Figure 2 (a) represents the offline phase, where a lap top is moved to different sample points in order to measure the strength of the signals that are emitted by the APs. These very signals are used to make fingerprinting maps of the area against to different APs. The results of these measurements are shown in Figure 2 (b), where we can see various signal strengths from AP1. In order to calculate location of the target node in the online position determination phase, the IPS, based on the fingerprinting maps of the area, uses the k-nearest-neighbours location algorithm 184.108.40.206. ProximityThe proximity location sensing technique calculates the objects’ location relative to an area or a known position. For this to be possible, the proximity location requires fixed number of detectors at the known positions. When a detector recognizes a tracked object, its position is calculated as the proximity area marked by the detector. As we can see in Figure 3, E2 and E3 are tracked objects. The dotted rectangle in Figure 3 represents the detector’s D proximity area. Even though E3 is outside the proximity area, both E2 and E3 objects are located by monitoring, which is the reason why proximity location sensing technique can not give absolute nor relative positions, as it is possible with other three techniques. This kind of location information is used for numerous location-based applications and services. For example, a proximity area represents a hospital room. Thus proximity sensing technique can accurately determine if a tracked object, in this case the patient, is in the room or not 220.127.116.11. Vision AnalysisAs shown in Figure 6, in the vision analysis, a location is calculated by one or multiple points 8. Furthermore, this type of positioning 9-12 delivers the efficiency and comfort to the users, as no additional tracked devices are required to be carried by the users. One or multiple cameras are commonly placed in the tracking area of an IPS, in order to cover the whole surface and take real-time images, from which tracked objects are identified. These observed images of tracked objects are scanned and searched in the pre-measured database, after which the position estimations are made. Lastly, vision positioning technique is capable of providing useful location context based on the 6 observed images. For instance, the vision technique can observe that a female person is sitting and using a laptop 2.2.2. Location AlgorithmsThe location algorithms calculate the position information of a object. For instance, in the triangulation technique example, an algorithm defines the location of the object after the distance between each reference point and target object is calculated. Researchers are developing different location algorithms in order to improve location calculation accuracy, which relies on whether the distance between each reference point and a target object, or some other location data, contains error or not. If this data contains a mixture of correct and erroneous data, then the algorithm tries to improve the accuracy, with the help of the a priory knowledge of an IPSs behaviors 2.3. Magnetic Positioning SystemIn this section we will introduce magnetic positioning system. The reason why magnetic signals are still used is that they do not suffer from line-of-sight problems, where an obstacle is blocking transmitter-receiver connection, and because magnetic signals offer high accuracy for positioning 2.MotionStar Wireless 13 is a motion tracking system, which simultaneously locates sensors in radius of 3 m area with the help of the pulsed DC magnetic fields. This is improved version of older MotionStar wired motion tracking system designed by Ascension Technology Corporation. The main difference between the MotionStar and MotionStar Wireless is that, in the newer version, there is no wire between a base station and a tracked person. Because of this feature, MotionStar Wireless system is used by different applications, such as virtual reality, biomechanics, animation, etc 2.As mentioned before, the MotionStar Wireless system locates sensors simultaneously in real-time. Current versions can track up to 120 sensors at the same time. The system consists of a base station, transmitter and controller, RF transmitters and mounted sensors. Body mounted sensors receive magnetic pulses from the transmitter and controller. RF transmitter, which is carried by the tracked person, is connected through wires to the sensors. Maximum number of connections in the same time is 20. Finally, when the base station receives the measured data from the RF transmitter, it calculates the position and orientation information of sensors and transfers this data via RS232 or Ethernet interface to the user’s computer. This data can be used for tracking and animation applications 2. 7 Body mounted sensors are small (2.54 cm × 2.54 cm × 2.03 cm) and weight 21 g, which makes them highly portable and comfortable for wearing. Sensors are connected to the RF transmitter (17.5 cm × 14 cm × 4.54 cm) that weights 0.99 kg via wires. The error range is about 1 cm for the static positioning. The major disadvantage is that magnetic trackers of the MotionStar systems are expensive. The battery life time is around 1 hour or 2 hours if used for continuous tracking, which is a short life period for daily use. Finally, as mentioned before, the coverage range of each transmitter is up to 3 m, which is not enough to be suitable for large indoor public services and applications 2.To summarize, the magnetic sensors are cheap, robust and small in size, which is favorable for indoor positioning estimations. Moreover, this type of positioning systems offers multi-position tracking and high accuracy. However, further researches and development are needed to increase the limited coverage range of the magnetic IPSs.