Motivation and goals
We have been hearing a lot about IPv6 implementation and the diversity it can bring to our lives. It has a very large address space which has been designed by keeping in mind the future needs.
The IPv6 technology was designed to allow a bigger number of addresses; improving the service quality and reducing the processing time on the routers (IPv6 does not fragment datagrams and neither check the integrity of the header -this happens at TCP level-).
Due to these specifications and as explained in chapter two, the headers of both protocols are different, both in size and in terms of field’s definition. The addressing of all elements involved in the communication is made through the IP network protocol so the datagrams which are sending to the Internet are different depending on the IP version used. The IPv6 datagrams are larger (since the header is bigger and there is not intermediate fragmentation) and are processed faster, so the performance of the network is different.
In recent years, large companies have not only to ensure that both, the network and the devices support IPv6 technology, but using this protocol does not consume a lot of resources. The operators have to know whether the protocol is operating correctly or not. For this purpose, an analysis should be made about the IPv6 protocol improvements on the functionalities borrowed from IPv4. Large telecom companies are asking the manufacturers to support Dual Stack functionality, which will allow made the network connection through IPv4 or IPv6. The aim of this master thesis project is to see how the most common and popular webpages and applications available on the market work through the mobile device. Thus, we will make several tests in order to compare the performance using different protocol versions.
Undoubtedly, the main service which we are going to test is if the webpages supports IPv6. This is quite interesting because most people use the device and data connection to access to the Internet cloud. Furthermore, it is also interesting to know if the most famous applications perform well over the IPv6 protocol.
Structure of the Thesis
To carry out and accomplish the objectives of this project, qualitative and quantitative methods have been used. The first parts of this master thesis project are based on a qualitative research methodology and then, with the information collected from the test on the laboratory, initiate a quantitative study.
The master thesis begins with a theoretical background study. This part includes the third mobile network (UMTS) architecture and the elements involved on the communication, especially within radio and core parts. We have mainly focused on the signalling and network protocols that we expect to use, including a deep analysis about internet protocol (IP). This section also includes a description of the UMTS network operation, linking the network elements explained on the architecture definition with the protocols responsible for each one of the operations.
Then, we discuss the laboratory configuration phase in order to evaluate the proposed performance parameters for both, webpages and apps. From the three options available to perform the testing (mathematical simulation, testbed or test over real network), we have discarded the mathematical simulation from the very beginning because it does not provide real information and the results are theoretical. Of the two remaining options, we have chosen the option of testbed which consists of a real network, but with controlled access to prevent interferences from other equipment that may appear in the real network. We also discuss what performance indicators the project will look at, why they are important, what kind of interactions or effects of input parameters we expect and how they are going to be evaluated.
The report continues with the evaluation methodology of the thesis based on several measurements taken on the laboratory to ensure the feasibility of the results, thus each test is going to be repeated many times. The results of the measured KPIs are presented as well as the explanation of the results. To end this chapter, we make a comparison between the expected results (before carry out the tests) with the real results obtained in the laboratory.
The final part of the project analyses the results collected on the laboratory, obtaining conclusions about the behaviour of the network, the devices, the operating systems, the webpages and the applications tested.
To finish the master thesis, the report raises possible lines of research in the future, which would serve as support for master thesis or PhD students who are planning to work on the UMTS performance over the internet protocol.
In this section, several previous studies conducted in areas relevant to this master’s thesis are summarized, mainly related to the performance of the TCP/IP stack over mobile networks.
The most of the previous work which have been carried out is related to the optimization of the TCP transport protocol to improve the throughput (amount of information flowing through a network according to the time). Mun Choon Chan and Ramachandran Ramjee  proposed and evaluated a mechanism called Window Regulator to improve the performance of TCP. They also demonstrated the efficiency of an algorithm for sharing the transmission and reception buffer improving the latency.
There are also too many studies about the TCP performance according to the congestion method used. Luca De Cicco and Saverio Mascolo checked  that TCP behaves similarly regardless of the congestion mechanism used.
Furthermore, studies on the IPv6 protocol have been conducted with special attention to the transition from the testbed towards the real network. Yi Wang, Ye and Xing Li Shaozhi analysed in 2005 several dual stack servers (dual stack means that support both IPv4 and IPv6) . Once evaluated the work, these problems are still present and the IPv6 protocol needs further optimization. The following table shows as main conclusion the high loss rate in IPv6, which can generate an increase in the number of signalling messages (having to resend lost packets again). These results are quite old (year 2005) and the problem of the high loss rate has been solved during last years since it has not been observed in any of the tests conducted in the laboratory.
Researchers from the University of Delft have also carried out a comparative analysis among protocols IPv4 – IPv6 at  from which can be concluded that the loading time and the delay is less in native IPv4 tunnels than in IPv6 tunnels, because the version 6 of the protocol is less optimized than the version 4 and therefore there is less number of optimal paths (see information about the IP tunnelling on the chapter two). In addition, researchers from the University of Budapest  have studied the methods of transition from IPv4 to IPv6 in the UMTS networks, discovering some limitations on non-native IPv6 tunnels. The main conclusion of these investigations is that we must stress the need for further studies aiming to help and urge the process towards the global native IPv6 coverage.
A previous study about the browsing optimization has been conducted as well by Binoy Chemmagate in Nokia networks . This research has analysed some HTTP and TCP issues on web browsing, testing alternative protocols as SPDY (that is a trademark of Google and is not an acronym) which provides header compression and multiplexing techniques. Using SPDY protocol we reduce the Page Load Time of a website, thus is highly recommended.
Another important published paper is about the performance improvement in the application and session layers . In general, we can observe that application and session layer techniques have a dominating effect in improving the web performance and commercial web servers and browsers should implement the HTTP-pipelining scheme, which provides noticeable benefits end-to-end user performance (in spite of the deployment of these services can be expensive).
The aim of this chapter is provide all the technical background to the reader in order to make easier the understanding of the project’s context. This chapter begins by introducing the third generation mobile networks (3G) and the signalling protocols involved on the communication.
To complete the theoretical framework, a general overview about the network protocols is presented. The IP protocol is defined, which is the most important in the research since the purpose of the master thesis is to evaluate it, as well as the evolution from IPv4 (used up to now) to IPv6. A brief introduction is also presented related the transport protocols and the application part.
The mobile communication system
The remarkable success of mobile communications systems of second generation (2G) defined on   and the needed to develop faster and more secure telecommunications services are the beginning of 3G.
To satisfy these demands, getting a similar quality than services offered by fixed networks and with a global perspective, emerges the mobile communication systems called third generation (3G).
The third generation mobile network
The 3G technology emerges to match the demanding needs of the users. Such technology is specified in the IMT-200 standardization defined by the ITU  and includes plenty of standards.
In Europe and Japan is standardized the UMTS system, which is based on the W-CDMA technology managed by the 3GPP  organization. The other mobile communications system which is regulated by the IMT-2000 specification is known with the name of CMDA2000. However, this technology will not be covered in this report since the whole Europe uses the UMTS standard. The main novelty of this technology, compared to earlier mobile networks, is that network architecture is divided into two domains, the UTRAN access part and the core network which is responsible for the interconnection to Internet or the public switched telephone network (PSTN).
The definition of the architecture is quite simple and intuitive, there are three distinct structures: the mobile station (MS), the radio network (UTRAN) and the core network (CN). Theoretical framework is widely explained in   using the specifications defined by 3GPP organism .
Importantly, this project only covers the study and analysis of the UMTS technology. Real networks, as seen from the operator’s side, are the result of a set of technologies such as GSM (2G), UMTS (3G) and LTE (4G) in order to satisfy all users. Therefore, the reader must keep in mind that the scheme depicted corresponds to a purely UMTS network and not to a current and real network operator one.
Table of contents :
1.2 Motivation and goals
1.3 Structure of the Thesis
1.4 Related work
2.1 The mobile communication system
2.1.1 The third generation mobile network
2.1.2 UMTS Architecture
The Mobile Station (Device)
The radio access network (RAN)
The core network (CN)
2.2 The signalling protocols
2.3 The TCP/IP protocol stack
2.3.1 The network layer protocols
IPv6 compared to IPv4
2.3.2 The transport layer protocols
2.3.3 The application layer protocols
2.4 UMTS performance on the internet network
2.4.1 Protocol operation in UMTS architecture
IP tunnelling call setup
TCP/IP stack operation
2.4.2 Web optimization for IP protocol in UMTS
3.1 Testbed environment
3.1.1 Steps of the testbed deployment
3.1.2 Problems and Issues solved
3.2 Scenario 1: Web Browsing
3.3 Scenario 2: Apps
3.4 KPIs measured
3.4.1 KPIS for browsers and apps
IP data and number of IP packets
3.4.2 Browser KPIs
Web Load time
3.4.3 Applications KPIs
3.5 Expected results
4.1 Browser Results
4.1.1 Data Volume
4.1.2 RNC Signalling
4.1.3 Web load time
4.2 Apps Results
4.2.1 Data Volume
4.2.2 RNC Signalling
4.2.3 Battery consumption
4.3 Analysis of the results
Conclusions and future work
5.2 Future work