Three Dimensional Localization of Work Pieces in Assembly Lines with Radio Frequency Identification

The knowledge of the location and the stock of work pieces are essential for companies to produce their products more effectively. Radio Frequency Identification (RFID) is becoming a widely used technology for this purpose. The goal of this thesis is to study the application of RFID for localization of work pieces in a manufacturing environment. To achieve this goal, this report reviews different localization and identification systems. Then, requirements for an ideal system are establish. The system that best meets the requirements is determined to be an RFID system. The components of an RFID system are described, and since the antenna is a critical component of RFID systems, different antenna constructions and the important Friis formula are discussed. After that, scenarios for localization of parts in an assembly line are listed. From these, the localization of work pieces in a warehouse is selected for further investigation. An experiment to determine the location of tags in this environment is designed. Data from multiple readings are recorded. Then the recorded data for the described experiment are analyzed. The data indicate that the read range of the antenna is not symmetric and that every tag has a different response in the amount of counts. All recorded data imply that the amount of counts depends on the orientation and distance between tag and antenna, the antenna power and the number of read tags. From the findings the localization of tags can be done by the following procedure. The first step is a scan with the maximum antenna power. The result of that scan is a list with all tags in the antenna read range. After this scan the antenna power has to decrease and the next scan starts. The decreasing antenna power leads to a decreasing read range. Below a certain antenna power, the tags read with the full antenna power cannot be read anymore because the distance between tag and antenna is larger than the antenna read range. With the known shape of the read range for the lowest antenna power were a tag could be detected it can be assumed that the tag is at the border of this read range. When this procedure is carried out at multiple antenna positions, the tag location could be determined.


INTRODUCTION
Today, efficiency is a key factor for every company to stay competitive in the global market. To reach high efficiency, resources need to be used without any waste.
These resources could be machines, employees, work pieces and raw material. The waste could be unused waiting time of a machine, the unused work capacity of human labor or a high inventory of work pieces and raw materials. Additionally, lost work pieces could result in multiple types for waste. First, lost work pieces might influence the working hours of a machine. In the case that the needed material is not distributed in time, the stock of the machine could run low which leads to idle processes. The idle process could also influence the entire system. Second, to find lost work pieces, employees are necessary, who search for the lost items. During that time the labor force of the employees cannot be used for processes with which the company would earn money. Finally, to compensate the lost material additional items may have to be manufactured, hence, more raw materials will be ordered and more machine hours and employees labor are necessary.
Therefore, the question which will be covered in this research study is how work pieces in an assembly line or factory can be localized to reduce the time to search for them if they are lost to prevent the company from wasting resources? The effects of the solution developed for the assembly line and the whole company could be an increase in the efficiency of the use of machines, labor and material by smarter distribution and logistics.
One possibility to find an answer to this research question is to focus on the development of a concept with a system which can localize the different targets. A issue is that just the positions of all work pieces would be known but not the identity of the items on these positions. To solve this issue, the information which might be transfer to the employees indicate, that a specific target is at a certain position. Therefore, the system has two tasks: localization and identification of objects.
The title "Three dimensional Localization of Work Pieces in Assembly lines with Radio Frequency Identification" could indicate that the technology was picked as the system to be used prior the beginning of the research. That was not the case. Therefore, in Chapter 2 different technologies for identification and localization are examined for how they fit the later identified requirements. Out of this process it turned out that the Radio Frequency Identification system is the technology to use.
The reason why Radio Frequency Technology is specified in the title is to find it in a shorter amount of time when research will be done in the future in the field of Radio Frequency Identification or localization.

AIM OF THE THESIS
Therefore, it is the aim of this research study to develop a concept for a system to localize and identify work pieces in an industrial setting. This includes how the data will be recorded and how the target position can be determent or calculated out of this data.

RESEARCH OUTLINE
To achieve this goal this study covers in Chapter 2 different identification and localization technologies as well as their advantages and disadvantages for the scenario of localizing and identifying work pieces. Subsequently, requirements for the desired system will be established. With the established requirements one system will be picked for further investigation. This is the Radio Frequency Identification System.
For this system the Friis formula, important parameter and possible designs will be discussed. Using this knowledge, an antenna will be designed, constructed and tested.
The measurements of the manufactured antenna and a commercial (Invengo) antenna will be compared to come up with the best system for the experiment described in Chapter 3 Methodology. Finally, the patents of the last 18 years for Radio Frequency Identification will be described to find out the leading companies and persons in this field.
In Chapter 3 scenarios where the tracking system could be an improvement will be developed. One of these scenarios will be picked for further investigation. After that, the experimental setup and the procedure will be described to record the data.
After this, Chapter 4 analyzes the recorded data. Here the strategy or the formula to determine the antenna position will also be developed. To calculate the position of the tags additional experiments are necessary, which are described, processed and their results analyzed.
Finally, Chapter 5 will sum up the findings and will give a brief description for the future.

REVIEW OF LITERATURE
For a system in a factory that can localize work pieces, it is also useful to identify these objects. With the information about the position and identity a computer system could calculate the most efficient way to retrieve these pieces, and give information to the employees where lost material or material for the next step can be found. First, this chapter will cover different identification technologies, and second, localization technologies. The technologies will be compared by their advantages and disadvantages for this scenario. With this information the ideal system will be picked.
Finally, the latest innovations for this system discussed in patents will be given.

IDENTIFICATION SYSTEMS
Automated identification of an object is very important in our modern world. The automated identification technologies enable a machine or device to read, collect and transfer data. The interaction of humans in the identification process should be kept to a minimum, to prevent human errors (Williams, 2014). For example, without barcodes the checking out process at a register in a grocery store would more time consuming.
Every single item would need to be identified by an employee and then the item number or price would be typed into the system. A system with no human interaction that can identify the object automatically would be much faster. In Kärkkäinen (2003) and Williams (2014)  contain about two kilobytes. In Figure 1 the different types of barcodes are displayed (Kärkkäinen, 2003;Williams, 2014).
The advantages of Barcodes are that they can be produced with a printer in large amount almost free. According to Williams (2014), a printed Barcode is cheap. In addition, the machine readability is good and the distance reading can be up to 50cm (Finkenzeller, 2010;Kärkkäinen, 2003).
A disadvantage of the identification with barcodes is that a so called "line of sight" between the tag and the scanner is needed to read the tag information. To read the two dimensions, more expensive readers are necessary. Additionally, the readability can be degraded by grease, grime and sunlight. Compared to other identification systems the data density is low and the reading time is slow at about 4 seconds (Finkenzeller, 2010;Kärkkäinen, 2003).

Radio-Frequency Identification:
For the identification of an object with Radio-Frequency Identification (RFID), a tag, a reader and an antenna are used. The tag must be attached to the target. A tag contains an antenna and a chip where the data is saved. Tags can be passive, with no included power source, or active, with an included power source (in the Figure 2-4 below a passive tag is visualized). The difference of active and passive tags is that active tags just need a signal from the reader, which starts the transmission process for the active tag. Contrary to active tags, passive tags generate the necessary energy to 7 power the chip and to send the information by induction from the reader signal.
Therefore, the read range of active tags is larger because the signal for passive tags cannot be sent over the same distance as the signal for active tags.
In some cases, software is necessary to run the reader. The software on a computer runs the reader and might display the returned information. The reader, which must be connected to the antenna, will send the signal to the antenna and the  (Leong, Ng, & Cole, 2006).
In the following paragraphs, the procedure is illustrated and explained. To read the tag information, first, a reader is necessary which generates the alternating radio frequency signal. The signal will be transmitted to the antenna and from there to the environment, as seen in Figure 2.

Reader Tag
Antennas of Reader and Tag chip Radio frequency signal send by reader The alternating radio frequency signal creates a current in the tag by induction in the tag's antenna ( Figure 3). In case of active tags, this current is the trigger to send the tag information, or in case of passive tags, the current has to power the process of sending the information.

Reader Tag chip
Radio frequency signal powers tag

Figure 3 RF-signal powers tag by induction in tag antenna: visualized from the information in the text
With this energy, the tag can send the saved data on the chip back to the reader.
The antenna of the reader receives the signal, transmit it to the reader and the reader decodes the antenna signal to machine-readable information, Figure 4. This information could be displayed on a monitor with a Graphical User Interface.

Reader Tag chip
Tag send data to reader The advantages of using a RFID system compared to a barcode system are that the line of sight is not required for the RFID system and the tags do not have to be oriented carefully. A RFID code can save up to eight kilobits and multiple tags can be read by one antenna. Additionally, grease, grime and sunlight affect the readability of a barcode, but the RFID system is not affected by these (Finkenzeller, 2010).
The disadvantage of RFID compared to the barcode is that the barcode is almost free, while the RFID tags cost about 12 cents per tag in units of one million.
Additionally, the batteries for active tags may have to be replaced after their life time (Want, 2006).
Due to these advantages, RFID system can be used in a lot of fields today, such as security, tracking, authenticity and electronic payments. In the security field RFID is used for access control in factories. Every employee receives a RFID tag in the form of a card or chip. A reader in front of every secure area can read the employees' tag number. This tag number can be compared on a computer with the numbers that have access. If the access data contains the employee number, the computer will send a signal and the door opens. Another security field is anti-theft. To prevent shop lifting tags are attached to the products. For this system the reader is installed at the entrances and exits of a building. If a product is purchased, the tag has to be removed from the product. If the tag is not removed, the readers will read the tag number and sound an alarm, preventing shop lifting. It is also possible to embed the tag inside of the product, say between the material layers of clothing. In this case the tag cannot be removed, so during the purchasing process the tag numbers of the purchased item have to be removed from the list of non-purchased products, otherwise the gate readers at the exits start the alarm. This system is also used in libraries, where a tag is inserted in the books (Want, 2006 RFID tags can also be used to make electronic payments. A car equipped with a tag could be read by readers at toll roads. With a list where the tag numbers are related to the information of the car owner's credit card, the billing can be done automatically. A similar system is already being used in the United States called E-ZPass. Another example is electronic tickets like ski passes. The tag number can be read at the entrance of the slope. The information will be compared with the numbers of the purchased tickets for that day or event (Want, 2006;RITBA, 213).

Bluetooth:
Objects can also be identified using Bluetooth technology. Bluetooth is a shortrange identification and communication technology between computers or mobile devices with radio links. Items in a factory can also be identified with Bluetooth chips.
The chip has to be attached to the item. The identification occurs by receiving the individual identification number from the chip at a reading device (Kärkkäinen, 2003).
The advantage of Bluetooth identification is that all mobile phones and computers, which can operate with Bluetooth, can identify the different chips. A conventional reader might not necessary (Kärkkäinen, 2003).
The disadvantages of Bluetooth are the necessary power supply for the chips.
Passive chips like passive RFID tags are not possible. The price of these chips in comparison to other systems is more expensive. One reason for the higher price is the underdevelopment of the Bluetooth standard and advances in mass production of the chips (Kärkkäinen, 2003).

Optical Character Recognition (OCR):
The OCR system was developed to read the human readable information on labels on the item. The readers for this system are much more complicated than readers of other identification systems (Finkenzeller, 2010), (Kärkkäinen, 2003).
The advantages of OCR identification are that the characters can be read by employees, so that the information read by the system can be supervised. Additionally, the density of the data can be very high and the labels can be cheaply produced with a printer (Finkenzeller, 2010).
The disadvantages of an OCR, system is that the systems are expensive and complicated. The reading speed is also slow and the maximum reading distance is very short. As with the barcode system, grime can influence the reading process (Finkenzeller, 2010).

Vision Recognition (VR):
All of the previous presented technologies require that a tag has to be attached to the object being identified. The Vision Recognition system creates an image from the compiled. The dimensions and shape of the object are used for identification by a software (Kärkkäinen, 2003).
The advantage of a VR system is that no labels or tags are necessary for the identification of the object (Kärkkäinen, 2003).
The disadvantage of a VR system is that an image is taken from the object. To take that image, the illumination conditions have to be very good for an errorless process. Due to these arguments VR is used in quality control and surveillance at assembly plants (Kärkkäinen, 2003).  Deak (2012) active and passive systems are described.
The defining characteristic for a passive system is that the target is not marked by a tag or other device, whereas an active system contains a tag or device attached to the target (Deak, 2012).
The localization of the target can follow the following procedure. The localization technology in use to send out a signal through all sources. These sources can be antennas or other transmitting devices. The targeted object will respond to the signal or the object tag will send the stored information back to the source. The sources receive the signals and transmit the information to a server. With the signal information, an algorithm on a computer can calculate the target's position using At RSSI a signal source transmits an outgoing signal. The target reflects the signal or sends information back. The received signal strength from the target at the signal source can be assigned to a specific distance between the systems signal source and the target. Due to the missing orientation information, the target can be anywhere around the source within the measured distance. Therefore, all possible target positions form a sphere with the signal source in the center (Bolic, Simplot-Ryl, & Stojmenovic, 2010). Like the RSSI system the TOA and TDOA calculate the distance between the antenna and the target tag, but in this case the time which the signal needs to travel the distance is recorded. With the known travel speed of the signal in a medium and the travel time, the distance can be calculated. Similar to the RSSI system the information of the orientation from the target to the signal source cannot be calculated from the signal information, so the target tag can be located on a sphere around the signal source.
In contrast to the prior three metrics, an AOA does not calculate the distance between the signal source and the tag. A system using AOA metrics returns the orientation of the direct line between tag and signal source related to an initial coordinate system. The tag can be anywhere on this line (Deak, 2012).
With the information of the metrics two different localization techniques can be used by the algorithm to calculate the target position.
First, is called trilateration. For this technique the distance information between target and signal source is used. Due to the unknown orientation of the target around the signal source, the distance information creates a sphere around the signal source.
Therefore, for the localization in a three-dimensional space, four signal sources are necessary; in a two dimensional space, three are needed. The positions of the signal sources need to be known. The advantage of this method is the fact that the target has to be on the spheres of all the signal sources which detected the tag. All of the spheres have to intersect at one point. Two intersecting spheres have a circular contact. In the case that they intersect at just one point, the target will be on the line of the shortest distance between these signal sources. The third sphere intersects this circle at two points and the last sphere is only intersecting with one of these two points. This point is the position of the target (Ko, 2010).
Trilateration for a two-dimensional case is displayed in Figure 5. In this case all possible target spots form a circle around the signal source. The calculation of the target location is similar to the three dimensional case. The circles of two signal sources intersect at two points. With the distance information of the third source, only one of these two points can be the target location. The trilateration can be used for RSSI-, TOA-and TDOA-systems (Ko, 2010). The second localization technique is called triangulation. The technique of triangulation is displayed in Figure 6 for the two-dimensional space. Similar to trilateration the positions of the signal sources (at least two) have to be known. For triangulation the orientation of the line between target and signal source is used. Therefore, one angle per source is necessary in a two-dimensional case, and two angles per source in a three-dimensional localization case. With the known angles  and , the unknown angle  can be calculated because the sum of all angles in a triangle is always 180 degrees ((Eq. 1) (Ko, 2010). In the previous paragraphs the procedure to calculate the target position from the returned information from the localization technology was described. However, in the 18 real-world, obstacles can be inbetween the reader and the target. These obstacles can lead to wrong information returns to the system. Therefore, the tracking of an indoor object is more challenging than the tracking of an outdoor object. The line of sight is not always possible in many cases. So, a requirement for an indoor localizing system is that the system has to be independent of the line of sight. Due to this requirement the Global Positioning System, for example, is not useful in the case of indoor localization because the line of sight is required (Deak, 2012). In the following paragraphs the hardware technologies used for indoor localization will be described, which metrics they use and their advantages and disadvantages for localization in an indoor assembly line.

Wireless Local Area Network (WLAN)
For device tracking with WLAN the target has to be equipped with a WLAN receiver. The receiver receives the strengths of the WLAN signal sent from the access point(s). The information of the signal strength can be assigned to the distance from receiver to a specific access point. The next step can be to send the information about the signal strength back to the access point. The calculation can be done through additional software running on a server or on a mobile phone or laptop using the locating software. Because of the described received signal strength for localization, the system is using RSSI with trilateration. Additionally, it is most suitable for tracking mobile devices like mobile phones or laptops because they are already using WLAN networks (Deak, 2012).
One of the advantages of localization with WLAN is the use of existing infrastructure. The access points are already installed, their positions are known and to add software to a mobile device or laptop is possible (Deak, 2012).
One disadvantage of the system might be that not significant access points are installed next to the assembly lines. Therefore, the position might not be exact enough for successful tracking (Deak, 2012).

Some examples of technologies which uses WLAN for localization is the Ekahau
Real Time Location System, the Microsoft RADAR or the Skyhook WLAN Positioning System (Deak, 2012).

Ultra-wideband (UWB)
Ultra-wideband (UWB) is not a localization system like WLAN or the other introduced systems. UWB is the technique to transmit radio waves with a high bandwith. The bandwith is the difference between the maximum and minimum used frequency. The difference is usually greater than 500 Megahertz (Mhz). A commercial system which uses an ultra-wideband signal is Ubisense. Ubisense uses their unique software, sensors and tags to run the system. The sensors return the TDOA and AOA, so the algorithm can triangulate the target position (Gezici, et al., 2005) (Deak, 2012).
The advantage of using an UWB signal is the bandwith. The different frequencies increase the chance that the signal can pass or transmit around obstacles. Another advantage of systems using UWB like Ubisense is that triangulation is possible, so just two signal sources are necessary to provide information about orientation.

20
The disadvantage of UWB systems is that they are under developed. A disadvantage for the Ubisense system specifically is that the price compared to other systems is high (López, Gómez, & Andrés, 2017).

RFID
Tracking with an RFID system works similar to localization with WLAN. In place of WLAN routers, RFID antennas have to be installed in the area where targets should be tracked. The targets have to be equipped with a RFID tag, which can send the identification number when the antennas send the trigger signal. The RFID antennas return the received tag identification numbers to the reader. The reader will transmit the information to the server with the calculation software to run the localization algorithm.
The basic idea behind tracking with RFID is that multiple antennas are mounted  Tag ID Antennas The relation between tag ID and antenna can also be established the other way around, where the antennas are listed with the corresponding tag ID's Table 3. Both tables above contain the same information but display it in a different way.
The advantage of doing this is that one to one relations between antenna and Tag ID can be identified easier. In Chapter 4 this advantage will be discussed in relation to the chosen scenario.
The RFID antennas return the tag identification numbers they receive. With this information and the known read range of the antennas, the location of an object can be 22 calculated. To calculate a more precise location, the signal strength could be related to a certain distance (RSSI). With the RSSI trilateration can be done.
Some RFID systems, like the Invengo system used later in Chapter 3, are able to return the RSSI. In Figure 7 the Graphical User Interface is displayed with the column for the RSSI (Invengo, 2015).
Column for RSSI value that the tags may not be triggered by the antennas (Condas, Devey, & Lemaire, 2010).
An advantage of tracking with RFID is the price for the tags. The price for large amount orders is about 13cents, when the tags are ordered in quantities of one million.
A disadvantage of tracking with RFID is that the orientation of the tag can have an influence of the tag detection. That leads to the issue that a tag can be in the read range of the antenna but is not triggered by the antenna signal. A solution to avoid this problem is to attach multiple tags at different orientations to the target (Condas, Devey, & Lemaire, 2010).

Infrared
Various systems have been developed which use infrared light for localization.
Some of the systems have the accuracy of a few centimeters, because they use TOA and AOA metrics, while others have room-level accuracy (Deak, 2012).
An advantage of localizing with infrared is that the technology is inexpensive (Deak, 2012).
The disadvantages of tracking with infrared are that the infrared signal does not travel through obstacles like walls. Therefore, the line of sight is necessary between reader and tag. In addition to this, the infrared signal can also be disturbed by direct sunlight and fluorescent lighting (Deak, 2012;Satyanarayanan, 2008).

Ultrasonic sound
Ultrasonic sound is also possible to use as the signal to transfer information to locate an object. The Active Bats system, developed by AT&T Cambridge, uses Ultrasonic sound. For this system, a badge is attached to the target. The badge sends the stored information about its identity. A receiver, usually installed every square meter in the ceiling, receives the signal and transmits the information to a computer which operates the localization information (Deak, 2012).
The advantage of the Active Bats system is that the calculated position is very precise. This precision is within just a few centimeters for over 94% of the readings (Deak, 2012).
One disadvantage of using Ultrasonic sound systems for localization like the Active Bats system is that that objects can reflect the signal. This reflection can cause 25 errors in the received data. To eliminate these errors and achieve the aforementioned accuracy, a statistical rejection algorithm is used. Another disadvantage is that the timeslots to transmit the signal are limited (Deak, 2012).

Bluetooth
Object tracing with Bluetooth works similarly to tracking with WLAN. A Bluetooth receiver detects the signal strength. For example, the BilpNet system uses the WLAN signal, which is transferred by Bluetooth to the mobile device. Then a program on the device can run the localization algorithm or the received signal strength is sent back and a server does the calculation (Deak, 2012).
The advantages of tracking with Bluetooth is that an already installed device is used. Additionally, the read range of the Bluetooth signal can be extended when additional antennas and amplifiers are installed (Deak, 2012).
The disadvantage is that the installation of additional antennas increases the costs of the whole system. Furthermore, every item, which does not have a Bluetooth receiver, needs to be equipped with one. As previously discussed in the introduction of the Identification Systems section, these receivers or tags are very expensive and underdeveloped (Deak, 2012).

REQUIREMENTS FOR THE SYSTEM
In this section, requirements will be established for an ideal identification and localization system in assembly lines. There are three issues which the system must deal with. First, obstacles like large objects in the line may reflect the signal. Second, a 26 signal strength decreases when it passes walls or obstacles. Third, other signals could interfere with the target signal. Out of these three issues the requirements will be established for a system which will be used for tracking and identification of work pieces in assembly lines. In Chapter 3 this system and its potential for use in different assembly line scenarios will be discussed (López, Gómez, & Andrés, 2017;Mehdipour, Trueman, Sebak, & Hoa, 2009).
The desired system should not only be used in developing assembly lines, but it should also be used as an improvement for existing assembly lines. In the case of an numbers. This leads to that the dimension in the scenario influence the read range.
All requirements are listed below again:  Line of sight is not required.
 Machines are not interfered, employees are not harmed.
 No effect of dirty layers on tag or item.
 No external power supply for tags.
 Artificial conditions like illumination is not necessary.
 Read range depends on the situation, should be larger than 1m.

SELECTION OF IDENTIFICATION AND LOCALIZATION SYSTEM
In this section an identification and localization technique will be picked for further investigation. Therefore, Table 4 is first established, which summarizes the requirements for an ideal system fulfilled by the identification technologies given at the beginning of the chapter. If a technology fulfills the requirement, it is marked with a checkmark (), if doesn't it is marked with an X. After that, Table 5 is established where all possible combinations of localization and identification systems are given.
The aim is to identify the best combination of identification and localization systems 28 because most systems cannot do both. Finally, one combination of systems will be picked for further investigation.
In Table 4 the identification systems and the requirements they fulfill are presented. Only RFID fulfills all requirements, while Bluetooth fulfills four out of the five requirements. Barcode and OCR have the same weaknesses in that the line of sight is required and unclean tags might not be readable. Additionally, OCR has a limited read range. Finally, Vision Recognition has the weakness that the item to be identified needs special lighting conditions and a line of sight.
Finally, the VR technology will not be used because the identification process with VR is problematic when the targets are stored in paper boxes or when the item is not illuminated appropriately.
lists all possible combinations of identification and localization techniques in a matrix. If a combination is feasible, the field is marked with a checkmark (), if it is not, it is marked with an X. Two techniques, RFID and Bluetooth, can do both identification and localization. Therefore, a combination of Bluetooth or RFID with another technique is not required, so the rows and columns of RFID and Bluetooth are marked with an X, accept the fields where the row and column of the two systems intersect.
First, the Barcode system is not suitable to be used for the desired system. One issue with the barcode system is that the line of sight between tag and scanner is 29 necessary, but in an assembly line this might not be possible. Another issue is that the barcode scanner may need to be aligned to face the barcode before scanning. This means that the position of the barcode is essential to know before scanning. This requires another system to locate the barcode.
Second, the use of an OCR system is not suitable as well because of the same issues the barcode system has (line of sight needed, effect of dirt). In addition, the OCR system has a short read range.
Finally, the VR technology will not be used because the identification process with VR is problematic when the targets are stored in paper boxes or when the item is not illuminated appropriately.

ANTENNA CALCULATION AND DESIGN
One key element of an RFID system is the antenna because the antenna must transmit the reader signal to the environment as well as to receive the tag answer.
30 Therefore, this chapter covers the most important antenna forms. After that the metrics gain, bandwidth, directivity and impedance were described. Finally, the Friis formula will be given, which calculates the antenna read range.

Dipole Antenna
First, is the dipole antenna. A dipole antenna can be build with two in line wires of the same length, displayed in Figure 9. In the middle of the two wires is an AC source, which can be the signal from the reader. The length  of the antenna is based on one wavelength of the frequency which the antenna should send. The wavelength can be calculated with (eq. 3). It is also possible to build an antenna with a shorter length. In this case the  can be divided by a nonnegative integer (Bevelacqua, 2009;Deavours, 2010).

31
Another way of designing a dipole antenna is the so called meandering dipole, shown in Figure 10. Here, the wire meanders, which leads to a more compact style of the antenna. The compact style can be important for the antenna to fit under labels (Deavours, 2010). The antenna will send electromagnetic waves when it is connected to an alternating current. The alternating current will charge the wires like capacitors positive or negative with the frequency to be sent. The moving electrons lead to an electromagnetic field around the wire. The electromagnetic field expend to the environment (Deavours, 2010).

Contacts for Signal
It is also possible to build monopole antennas. A monopole antenna containing just one of the wires to send out the radio waves, displayed in Figure 11

32
There are multiple ways to build an antenna. Additional antenna shapes are presented at Bevelacqua (2009). In the next paragraphs some important antenna measurements and the Friis formula will be discribed.

Bandwidth:
The bandwidth of an antenna is the difference between the highest and lowest frequency which the antenna can send (Bevelacqua, 2009).

Impedance:
The impedance is the relation of the current and the voltage. It can be a complex number, if current and voltage are not in phase (Bevelacqua, 2009).

Radiation Pattern:
The radiation pattern is the information how much power is transmitted by into ever direction around the antenna. Therefore, the shape is three dimensional. For a better visualization a two dimensional charts present layers of the shape of the three dimensional one. The chart layers are the so called E-and H-plane (Bevelacqua, 2009).
On the chart of the radiation pattern, the half power beam bandwidth is the angle between the two lines which can be drawn between the center of the antenna and the 33 points at the main lobe where the power of the main lobe decreased by 50% or 3 dB (Bevelacqua, 2009).

Gain:
The gain describes how much power of the antenna is transmitted in the peak direction compared to an isotropic antenna.
With the Friis formula the power which a tag receives can be calculate The formula indicates that the power which the tag receives depends on, the distance between antenna and tag, the antenna gain of antenna and tag, the wavelength 34 of the signal and on the power which was sent out by the antenna. With the known minimal power to operate the tag, the formula could be converted to calculate the read range of the system (Bevelacqua, 2009).

Antenna manufacturing
With the knowledge of the last paragraphs a monopole antenna will be developed and tested in this section.
For the monopole antenna the cupper strips from an electric cable were used. The isolation was removed and the wires were interweaved together. This copper wire was then cut to the length of 16.5 cm, which is the half of the wavelength of the UHF signal. For the connection to the Invengo reader signal (is described in Chapter 3), the plastic cover from the antenna was removed. Then the metal antenna in Figure 12 was replaced by the copper string ( Figure 13). The electric connection to the signal was ensured by drilling the screw through the one end of the copper string. Then the tag was moved from a point outside of the read range into the read range.
The result is that the antenna has a maximum read range of less than one meter.
With a read range of less than one meter the maximum possible reading distance of the manufactured antenna is below the read range of the Invengo antenna which is about 6 meters (data in Chapter 4).
Reasons for this result might be that the wire is maybe not cut at the best length which could have an effect on the signal frequency. Another reason might be that the Invengo antenna has a larger surface and contains more material, which is maybe optimized for the reader signal, than the copper wire.

PATENT RESEARCH
To get an overview about the leading companies which provide RFID localization systems, this section will sum up patents from 1 st , January 2000 to 31 st , May 2018.
The aim is to get an idea which companies or persons are leading the market. The 36 claim of the patents is not being described. The patents were found on Google Patents by searching for "localization RFID" and "RFID localization".
When using "localization RFID" 16 patents were found while for "RFID localization" 54 patents occurred. Out of these 70 patents 21 patents were selected because these patents covered the localization of a target like robots or to build up a warehouse management system. The list with the selected patents is given in Table 6.

Some inventors are written in Mandarin or Korean letters
Most of the patents from Table 6 were invented by different persons. Just two times the inventor of patents were the same. These were the patent US7619532B2 and

METHODOLOGY
In this chapter scenarios will be described, where localization with RFID can be useful for an assembly line. For each scenario, obstacles and how the RFID system can be used to avoid them are described. Subsequently, one scenario will be selected for further investigation. The chosen scenario will be set up in an experiment and the data will be recorded.

SCENARIOS
Before localization scenarios can be named, the flux of the work pieces or the raw material is important to identify. The steps of the process to transform the raw material into a finished product are shown in Figure 14. A process may start with the delivery of raw material, like steel coils. The raw material could be stored before it is delivered to the machines to produce work pieces. The work pieces produced could be stored again or delivered to the assembly lines. The assembled products can be stored or dispatched immediately to the customer.
From this material process description, the following stages can be identified:  In the following paragraphs, scenarios are described for the identified stages. The aim is to describe the obstacles in each step and how they could be solved with an RFID localization system.

Scenario for Delivery/Dispatch:
Delivery is the entrance of the raw material or components into the factory and dispatch is the exit where the products leave the company. During delivery situations, 42 a lot of material may arrive at the same time. Even though this may be a hectic moment for the employees where errors can happen easily, all information about the received items need to be inserted into the inventory system without error. Dispatch is the stage of the process where the company has possession of the products for the last time and where the company can make sure that the outgoing dispatch is correct. At both stages, automatic scanning of the items could be done by RFID readers and antennas with tags attached to the items.
All processes described for the delivery and dispatch stages involve humans. Therefore, a system which does not need employees will help to reduce human errors.
Despite to the automatic process, some human oversight might be necessary for monitoring.
A sketch of a possible solution is displayed in Figure 15. RFID tags are attached to all the boxes which contain the products. When the boxes pass a gate (checkpoint at the delivery/dispatch, were tags are recorded) with antennas, the antennas will read the tag numbers. These tag numbers could be compared with the tag numbers of the products which must be delivered or dispatched. With this information the stock system could automatically be updated,

Scenario for Machine line up
A machine which does not work due to a lack of material stock does not earn money for the company. A system which notices low stock of materials may prevent this situation. To obtain the necessary information for the system in this scenario, an antenna could be installed in a position where it is able to scan the whole input line of the machine as shown in Figure 16. With the information about the items in the input line, it could also be ensured that the right items were delivered.
An advantage, of such a system might be that the system could order the necessary work pieces in advance before they run low. With all the antennas in the factory, the track of the products could be recorded.

Scenario for Area Localization
Another scenario occurs when items must be located in a production facility. In this case antennas mounted on the ceiling could create different reading zones, like those displayed in Figure 8. With information about the zone where the item is located the product could be found faster. If the RFID system in use can record the RSSI, the location of the target might be calculated more precisely.

Scenario Storage
In storage a lot of products could be placed in a very small amount of space. To lose track of inventory would cost employees time to find them again and cost the company money. This issue can lead to products which are not delivered on time because their location is unknown.
In this scenario it is assumed that the items are stored on pallets, because these pallets make the logistics of moving the items easy by using devices like forklifts.
In this scenario, it can be assumed that a beam can move between storage shelves, Due to the large number of stored work pieces per area, this scenario will be chosen to investigate the advantages of RFID to minimize human error by an experiment.

EXPERIMENT
In the following paragraphs the experiment will be explained. The experimental setup will be described, as well as the process of data recording. The data will be analyzed in Chapter 4.

Setup of the Experiment
The experimental setup must reproduce the conditions corresponding to the situation shown in Figure 17. In Figure 17 all the tags are in one plane. For the case that all tags are in a vertical plane (that is the case in the Figure 17), a frame on which the tags would be mounted, might be necessary for the experiment. This vertical plane could be changed to a horizontal orientation for the experiment, the advantage being that no frames would be needed. Another advantage for this case is that using tags in a horizontal plane is much more easily accessible because they could be laid on the ground. Therefore, the position of the tags for the experiment will be on a horizontal plane.
Before the experiment was set up, the readability of tags laying on the ground was tested. The result was that tags on the ground could not be read by the antenna. An explanation for this result could be that the concrete floor contained steel, which affected the signal. However, this hypothesis could not be proved. A solution to evade this obstacle could be to put them on a table with a wooden tabletop. The thickness of the wooden tabletops used in the experiment (see an image of the tables in Figure 19) was large enough that the steel in the floor and of the table's frame did not influence the tag's readability.
To transform the situation of Figure 17  In the experiment 19 tags were used to simulate the situation of an empty slot to discover how it might impact the data. Figure 19 and Figure Table 7 the tag identification numbers are related to the tag positions in Figure   21. Only the last four digits of the tag ID are given because they are unique for every tag.
The tags placed on D, F and T did not return their identification number. In the column on the right the return value is given. The tag on position T returned an empty identification number (Table 7). To record comparable data for every trial, the tags were never moved from their position.
To place the antenna in the correct orientation above the tag for scanning, a movable table was used. On the table a cardboard box and a plastic crate were used to boost up the antenna on a beam so that it could scan above the tags. On top of the box and crate a cardboard beam was mounted. The antenna was fixed at the end of the beam ( Figure 22).

Figure 22 Construction of the movable Antenna
The orientation of the antenna is important for the experiment because the readability of the tags depends on its orientation. The orientation of the antenna with the main lobe facing the tag below it was chosen because it ensures that the tag below is in the read range ( Figure 22). Another advantage is that in this orientation the experiment is simulating how the antenna could pass through the space between two shelves, which likely is very small.
For the experiment, the Invengo system containing a reader, an antenna and software were used; the tags used were from Omni-ID. The reader was a XC-RF861 with four possible antenna connections and the antenna was a XC-AF12-A. To operate the system, the RFID Reader Gereric Demo Software in the Version of 1.2.3 was used.
The tags that were used were Omni-ID Power 415.
The XC-RF861 reader is shown in Figure 23 and its model label in Figure

Process of the experiment
In the following paragraphs the process of the experiment and data recording will be described. Each scan at each tag position with each power level was done for 60 seconds (1 second). After that, the recorded data was saved, and the power was lowered to the next increment. The recorded data will be analyzed in the next chapter. According to the data in Appendix A, decreasing the antenna's power leads to a decrease in the read range. With this smaller read range, the localization of the tags can be done. Therefore, the antenna should be set to the maximum power that allows just one tag to be read with a total count of about 3000 responses. With this result, the responsive tag is most likely at the scanned position for the set up in the scenario because all tags which lay below the antenna responded with a rate of about 50 readings per second. Tags which were not directly under the antenna responded with a far lower rate. Another consequence of decreasing the antenna's power, which can be seen from the data, is that the shape of the read ranges shrinks. It only occurred once (Position E from 30 to 29 dB) that a Tag ID was read at a lower antenna power level, which could not be read at a higher antenna power.
The scanning of position S is a special case. On position S, no tag was placed to simulate the situation of a free space or a broken tag. However, with the information from the paragraph above (one tag read with 3,000 responses) it can be determined that position S is the position with no tag. This is because the tag 1662 was read with a maximum number of responses of about 2,500, while it was read more than 3,000 times on position T. This implies that the read tag must be at another position (position T) and position S is empty. Figure 32 visualizes another point about read ranges, which is that they do not always have the shape of a circle. The scanning of position F shows that the tag on position N was scanned, while the tags on positions H and M were not scanned, though their distance to the antenna is shorter. This leads to the hypothesis that the tags also have an influence on their readability and, therefore, on the read range. To prove this hypothesis, the following second experiment was carried out. The aim of this second experiment is to determine whether the selected tags respond equally or not to the same conditions. It was not carried out to describe the level of response at those conditions.

Second Experiment (Tag Responsive Pattern Comparison)
For the second experiment, two tags-one with a high level of response and one with a low level of response-were selected. The tag on position N was selected as the tag with the high level of response because the tag was readable even with an antenna power of 27 dB on position F. On the other hand, the tag on position A was selected as the tag with the low level of response because this tag was only readable when the antenna was directly above the tag. The setup of the experiment is visualized in Figure   33.    At every scanning position, the antenna was facing the experimental area ( Figure   39).

Figure 39 Antenna at position X1 facing tags
To place the antenna with the right distance to the tags for the storage scenario, it was assumed that the pallets with the tags on the Positions A, B, C, D and E were 67 standing on the floor. Therefore, the antenna was placed right next to the tag at the scanning positions U1, U2, U3, U4 and U5 (Figure 40).

Figure 40 Antenna at position U2 facing tags
For the scanning positions X1, X2, X3, X4, V1, V2, V3 and V4 the antenna was not placed right next to the tag. For these positions the width of the pallets is important. In the scenario the width of the pallets was 1.2 meter and the tags were mounted in the middle of the pallets. Therefore, the distance from antenna to tag was set to 0.6 meter. This is visualized for the Position X1 in Figure 39.
The same procedure was done for the positions W1, W2, W3 and W4. For these scanning positions the antenna was placed at the considered height of the pallet, 2.5 meter (pallet measurement information is displayed in Figure 18), Figure 41. Just the reading at the positions W4 and W5 scanned the faced tag at the end. In this case the total counts were about 2300 and 2200. This indicates that recorded data were just one tag was read with a total count of about 2200, the tag in front of the antenna was scanned, while a total count of 1500 or less indicates that the antenna is not facing the read tag.
One hypothesis for the differences to the other scanned data at the other positions is that the distance is larger. Additionally, the orientation, of the tags is different at the reading positions, but may have an influence. At the scanning positions W and U, the tags faced the antenna with the long side, at the positions V and X with the short side.
To proof this hypothesis a similar experiment to the one displayed in Figure 33 and The tag was moved to the position were the sharp drop in the read ranges appeared. Then the distance between tag and antenna was measured. The measured data are presented in Appendix E and visualized in Figure 46.
The area 90 to 270 degrees (clockwise) is the area at the back of the antenna were also the screws to mount the antenna and the attachment for the reader signal are. In this area the read range is low. Against to this is the area from 270 to 90 degrees (clockwise) which has a wider read range with a peak of about 6 meters at 0 degrees.
Zero degrees was also the orientation for the tags in the third experiment.

CONCLUSION
In this chapter the findings of Chapter 4 will be summarized, a method to locate the tags for the selected scenario of Chapter 3 will be given and a view to the future will be given.

Findings
In the previous chapters 3 and 4 the experimental conditions were given and the recorded data were analyzed. From the recorded data for the experiment described in Chapter 3 it can be seen that the real read range of the antenna, shown for the positions B, H and F in Figure 30, Figure 31 and Figure 32, is not a circle as it was assumed in Figure 8. Therefore the concept of dividing the area in different zones, which are characterized by the antennas which could detect a target in this zone, is not possible because the read ranges have, in addition to the non circular read range, different shapes. So, another method is needed determine the tag positions out of the recorded data. One possibility is to start the scanning with the highest antenna power and decrease the power by 1 dB after every scanning. The consequence is that the read range will shrink. This procedure needs to be done until just one tag can be read with a total count of 3,000 in one minute. The procedure ensures that positions were no tag is placed will be found because at these empty positions the total count will be below 3,000 readings in one minute due to the greater distance between tag and antenna.

77
Another assumption which can be made out of the recorded data is that the tags may respond differently, depending on the orientation of antenna and tag and the distance between both. Therefore, the second experiment was done. The analyzed data indicate that the compared tags are responding on a specific amount of counts for the full antenna power. At a certain lower antenna power the number of counts read per second decreases sharply. The power where this sharp decrease occurs is different for all measurements depending on the distance between antenna and tag, the tag itself and the tag and antenna orientation. The result is that the responsive behavior of every tag has to take into consideration as well.
After that, the tags were also scanned from the side, Figure 38, to investigate if this antenna to tag orientation would be suitable to calculate the tag position with the previous explained concept of reading zones. The results show that the read ranges in this setting are, as well as for the first presented experiment, not of a circular shape.
The effect that a decreasing antenna power decreases the readable range was the same.
Additionally, the response of a tag was recorded for various antenna power levels, antenna to tag distances and for the orientations that the long-or the short tag side is facing the antenna. The results of this experiment are that the number of responses for a tag which is facing the antenna with the long side is equal or higher than for the same tag which is facing the antenna with the short side. Additionally, for the orientation of the long side, the necessary antenna power to trigger the tag is less than for the orientation of the short side.
To understand the behavior and the influence of the antenna, the read range of the antenna was measured at different angles. One of these measurements is shown in Figure 45. Therefore, a tag was moved on a string, which was orientated at the desired angle, to the unknown maximum antenna read range. When the maximum read range for the antenna at a certain angle was found, the distance between antenna and tag was measured. All measurements are shown in Figure 46. Figure 46 indicate that the read range of the antenna is not a circle and is not symmetric.

Localization method
Despite to the found obstacles that the read range of the antenna is not a symmetric circle, the location of the tags in the picked scenario of Chapter 3 can be determined by the following procedure. The first scan should be done with the full antenna power. The reason is to ensure that tags are in the antenna's read range. If no tags are detected, the antenna can move on to the next position because with decreasing antenna power the read range would also decrease. For the case that tags are detected the antenna power has to decrease until a sharp decrease in the total counts can be recorded. For this situation the location can be estimated with the used antenna power.

Future view
With the findings it indicates that the positions of the tags can be determined in the covered scenario. However, not all possibilities of the previous chapters were discussed in detail. As an example further investigation can be done for the described but not selected scenarios of Chapter 3. Furthermore, research could be done to optimize the antenna as well as the used tags for the desired case.

79
Another possible research field is to establish and process experiments to describe the interaction between the antenna and tag better. Therefore, the responses with many more tags than just two tags would needed to be recorded. When the Invengo system will be used for further investigation, the purchase of the complete version could be important to receive the RSSI. With the RSSI the distance between the tag and antenna could be determined. To determine the distance multiple tag readings might be necessary.