Objective assessment of IP video calls with Asterisk

INFORMATION AND COMMUNICATION TECHNOLOGIES AND SERVICES VOLUME: 10 | NUMBER: 4 | 2012 | SPECIAL ISSUE 246 © 2012 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING OBJECTIVE ASSESSMENT OF IP VIDEO CALLS WITH ASTERISK Lukas KAPICAK1, Pavel NEVLUD1, Martin MIKULEC1, Jaroslav ZDRALEK1 1Department of Telecommunications, Faculty of Electrical Engineering and Computer Science, VSB-Technical University of Ostrava, 17. listopadu 15, 708 33 Ostrava-Poruba, Czech Republic [email protected], [email protected]z, martin.m[email protected], jaroslav.z[email protected] Abstract. The paper deals with an objective assessment of IP video calls transmission over GSM and UMTS networks. Video transmission is affected by many factors in mobile network. Among these factors belong packet loss, latency and transmission rate of the mobile network. Network properties were simulated by Simena network simulator. Our team have developed a unique technique for finding defects in video appearing in video calls. This technique is built on modified Asterisk SW PBX with enabled video recording and playback functions. Transmitted video files are compared with original video file by means of size of transmitted video file and invideo-defects. We are using MSU VQMT software for finding in video defects; more precisely we are using VQM method for comparing two video sources. Keywords Asterisk, comparison video files, H.263, mobile networks, video calls. 1. Introduction The members of our team participate on the INDECT project [1]. They are responsible for developing the IVAS (Interactive Video Audio System) system [2]. The IVAS system represents multimedia gateway between the INDECT portal and end-user equipment. End-user devices are mainly mobile devices with support for VoIP [3], [11] communication through data transmission via mobile networks. All multimedia functions are ensured by IVR (Interactive Voice Response) and IVVR (Interactive Voice Video Response) functionality [4]. The transmission medium is mostly represented by mobile networks. We tested all available mobile data transmission techniques but only networks with higher transmission rate than offer EDGE (Enhanced Data rates for GSM Evolution) have sufficient characteristics to transfer video calls over IP. Data transmission rates in mobile networks has limits in transmission speed, latency and packet loss. It is very important to find these limits for transmitted video content. We were using Asterisk as a multimedia gateway. Asterisk provides basic functionalities such as recording and playback video files, generating IVR and IVVR menu etc. We decided to use recording and playback functionalities to assessment video files. We used two Asterisks servers for this purpose. First of them establishes video call and played pre-prepared video file and the second Asterisk records that video file. We decided to simulate mobile networks characteristics. Transmission rate in mobile networks is very variable parameter and it is affected by many variable factors such as the number of connected users, weather, available network transmission rate etc. We used Simena network simulator [5] to simulate mobile network characteristics. The network characteristics came out from multiple measurements in tested mobile networks. Saved video has to be analyzed with objective method. We have found project MSU VQMT (MSU Video Quality Measurement Tool) [8] specialized in comparison differences between original and encoded video file. MSU VQMT has many plugins and we have chosen plugin called VQM [9]. With this plugin, we were able to probe saved video files and compare it with original video file. 2. Testing Configuration As was mentioned before we used Asterisk for establishing audio and video call. In the default configuration, Asterisk does not support video recording and playback so Asterisk needs to be upgraded with plugins which enable recording and playback video files. We implemented into Asterisk app_mp4.c plugin, which extended Asterisk for new functions – mp4play and mp4save. These two plugins allowed us to record and play video files supported by an Asterisk and by mobile devices [6]. The extensions.conf needed modification to record video files. It was necessary to prepare two Asterisk servers. First Asterisk established INFORMATION AND COMMUNICATION TECHNOLOGIES AND SERVICES VOLUME: 10 | NUMBER: 4 | 2012 | SPECIAL ISSUE © 2012 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 247 video call and second Asterisk received the video call and then saved the video call file into predefined directory. First Asterisk server had to have appropriate configurations in sip.conf configuration file to forward video call to the second Asterisk server. There was a network emulator Simena [5] located between Asterisk servers. Connection parameters were adjusted according to measured parameters in mobile networks. We measured CSD (Circuit Switched Data), HSCSD (High-Speed Circuit-Switched Data), GPRS (General Packet Radio Service), EDGE and UMTS (Universal Mobile Telecommunications System) parameters [6]. We did not measure HSPA (High Speed Packet Access) or HSPA+ network parameters. Table one represents measured transmission speeds. We have chosen transmission rates about 214 kbps and 1990 kbps. Mobile network technologies such as CSD, HSCSD, GPRS and EDGE does not provide sufficient transmission rate for real time video transmissions. Encoded video file was pre-prepared with these parameters:  video resolution 176x144 pixels,  25 frames per second,  video codec H.263 with 304 kbps data rate. Tab.1: Measured transmission speeds in different mobile networks [7]. Transmission technology EDGE UMTS FDD UMTS TDD HSDPA R5 Downlink [kbps] 126,57 123,19 214,77 1996,91 Uplink [kbps] 101,35 113,59 157,56 330,45 Latency [ms] 193 144 124 87 These video parameters depict that transmission speed 214 kbps will not fully cover needs of video transmission. We have chosen video data rate in case of comparison defects in picture between transmission speed 214 kbps and 1990 kbps. We set these transmission speeds in Simena network simulator and we were changing packet loss and latency. Testing architecture is displayed in the figure one. Fig. 1: Testing architecture with Simena network emulator. 3. Video File Size Comparison Our method allows us to save video call into predefined storage in Asterisk 2. Video has the same parameters as a video call so we can simply compare file size of these video files. Table two and three depicts differences in video file size. It is noticeable that rising latency has not influenced on the video file size. On the other side, rising packet loss and transmission rate has influence on the video file size. Figure 3 and Fig. 4 depict decreasing character in dependence on ascending character of the packet loss. Tab.2: Measured transmission speeds in different mobile networks [7]. H.263 214 kbps 1990 kbps Latency [ms] Size [B] Size [B] 0 748112 991129 50 837544 991129 150 743340 991129 250 711935 991129 350 688161 991129 500 962987 991129 Tab.3: File size of saved video files – packet loss. H.263 214 kbps 1990 kbps Packet Loss [%] Size [B] Size [B] 5 793403 944177 15 672654 791849 25 757362 746104 35 656917 661522 45 560021 548463 55 379394 353694 Fig. 2: Graph with packet loss with transmission rate about 214 kbps light blue – original file size, dark blue – transferred video file size). It is noticeable to see differences between file size in figure two and three and in table two. When we analyzed packet loss between data rate 214 kbps and 1990 kbps we have found similar video file size in packet INFORMATION AND COMMUNICATION TECHNOLOGIES AND SERVICES VOLUME: 10 | NUMBER: 4 | 2012 | SPECIAL ISSUE 248 © 2012 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING loss 25 %. Transmission rate was different in these two cases so it was necessary to analyze these two situations. Fig. 3: Graph with packet loss with link rate 1990 kb/s (light blue – original file size, dark blue – transferred video file size). 4. Analysis of the Video Content Saved video files should be analyzed by the objective method. We used project MSU Video Quality Measurement Tool [8]. This software allows us to compare one or more video files. It is designed for analyzing differences in encoded video files but we used this MSU VQMT for analyzing differences between original and transmitted video file. MSU VQMT contains different tools for analyzing video files. Fig. 4: Overview of Video Quality Measurement [9]. Fig. 5: Differences in the picture with different VQM – original/VQM = 15,5234. We used DCT – based Video Quality Evaluation tool (VQM) [9]. This method is used for analyzing defects in compressed video files. DCT-based video quality metric (VQM) is based on Watson’s proposal [10], which exploits the property of visual perception. This metric uses the existing DCT coefficients, so it only incurs slightly more computation overhead [9]. We had four sample groups with saved video files. Two sample groups were affected by latency in video transmission. First group was affected by low transmission rate too. Table 4 and 5 show number of frames in saved video files. Number of frames in the original video file was 391 frames. The MSU VQMT has implemented technology (by extension) to optical comparison original video file with transmitted video file (saved video call file). With this extension, we were able to compare every single frame of the video file. Table 4 shows information about VQM values depending on packet loss. In the case of that packet loss has zero value, the VQM value is very low. We have found that the main differences in video quality are at the beginning of the video transmission and at the end of the video transmission. Average VQM value has not rising character. It is noticeable in Tab. 4. Tab.4: VQM values depending on the packet loss. H.263 214 kbps 214 kbps 1990 kbps 1990 kbps Packet Loss (%) Frames Avg. VQM Frames Avg. VQM 0 255 0,02205 391 0,01354 5 297 12,87321 366 13,13988 15 266 13,45568 319 14,22181 25 297 14,83738 298 14,42438 35 260 14,28836 263 14,95820 45 225 15,11789 230 14,90110 With rising packet loss rises frame dropping too. From a statistical point of view, there are in-video defects spread out between totally dropped frames and between damaged frames. Tab.5: VQM values depending on the transmission latency. H.263 214 kbps 214 kbps 1990 kbps 1990 kbps Latency (ms) Frames Avg. VQM Frames Avg. VQM 0 255 0,02205 391 0,01354 50 305 2,17959 391 0,01354 150 253 0,02214 391 0,01354 250 241 1,31047 391 0,01353 350 226 0,03398 391 0,01354 500 374 0,46453 391 0,01353 Table 5 depicts VQM values depending on the transmission latency. The VQM values are independent on the video latency. Low frame number is caused by low level transmission rate (214 kbps). In case of the situation, when transmission rate is enough to cover needs of video call, the VQM value is nearly zero. There are no differences between original video file and transmitted video file. INFORMATION AND COMMUNICATION TECHNOLOGIES AND SERVICES VOLUME: 10 | NUMBER: 4 | 2012 | SPECIAL ISSUE © 2012 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 249 Fig. 6: Comparison graph with VQM values for 214 kbps data rate – transmission affected by latency 150 ms (blue) and 25 % packet loss (red). Figure 6 displays differences in VQM value. The red graph shows very variable characteristics of the VQM value. The differences between original video file and transmitted are significant. On the other hand, the VQM transmission values are influenced by the latency nearby zero values. In some circumstances, the VQM value could reach a larger value but this situation can occur only at the beginning of video call transmission or at the end of the video call. Low transmission rate does not cause changing of VQM values. Low transmission changes only number of saved frames and it causes reducing of saved video file too. Fig. 7: Comparison graph with VQM values for 214 kbps (blue) and 1990 kbps (red) data rate – transmission affected by 25 % packet loss. Figure 7 depicts VQM values character for transmission rate 214 kbps and 1990 kbps. The character of the curve is similar for both values (blue for 214 kbps and red for 1990 kbps). The highest VQM values are at the beginning and at the end of the video call. 5. Application Results from these tests could be used for both video and performance benchmarking. It can describe server performance for a specific number of video calls. We can also describe transmission line performance and we can also define the number of successfully transmitted video calls at one time. Transmission errors could be divided into these categories:  Video call file has the same size such as the original video file. Transmission line covers transmission demands of the video calls. Transmission delay caused by latency is not found.  Video call has the smaller size compared with original video file and there are small VQM values. The transmission line does not cover demands of the video calls. Transmission delay caused by latency is not found.  Video call has smaller size compared with original video file and there are high VQM values. All broadcasted packets do not achieve destination and the packets are lost during transmission. Video contains many in-video defects. With these results, we can describe link properties and also define the cause of transmission errors. This approach allows to define network properties according to video calls characteristics. 6. Conclusion Video transmission is very important part of the IVAS system. Many other services such as video call, online streaming or IVVR technology depend on video transmission. The IVAS system is designed for mobile end users, so it is necessary to count with mobile data transmission. End user devices are represented by smartphones or tablets. We tested transmission of video calls in mobile networks. These networks have some limitations in transmission rate, latency and packet loss. Measured parameters have very variable pattern so we decided to simulate these parameters with Simena network emulator. Simena network emulator allowed us to configure virtual network parameters on the wired transmission line. Testing environment consisted of two Asterisk servers and Simena network emulator. Asterisk server was equipped by module which allowed record and play video files saved on the local storage. Video call was established through this simulated network by specific CLI commands. The thirst tested parameter was the size of the saved video call. Video file size depends on two parameters – transmission rate and packet loss. High latency does not influence the size of the video file. We decided to investigate differences in video content. There are two ways how to analyse video content – subjective and objective analysis. An objective method has been chosen for our tests. MSU VQMT provides needed tool for video analysis. This software includes many plugins but we have chosen only VQM method. With this method, we were allowed to analyse differences and errors in saved video files. We were able to find specific errors and then we divided in-video errors into specific categories defined in this paper. According to these errors we described transmission line and found a solution to improve network ability to transfer video calls. Testing network consisted of two Asterisk servers can be used for testing any IP transmission line with video calls. It is possible to establish benchmark and INFORMATION AND COMMUNICATION TECHNOLOGIES AND SERVICES VOLUME: 10 | NUMBER: 4 | 2012 | SPECIAL ISSUE 250 © 2012 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING security [12] tests with appropriate configurations in Asterisk dial plan. Results from the tests could be used for new tests. The solution is almost ready for system benchmarking. We can prepare scripts for establishing larger number of video calls in one moment. The line between two Asterisks could be wired or wireless. We are planning to test our solution in real mobile network too. We plan upgrade this solution with development of new Asterisk module which will combine technics comes from MSU VQMT. Testing video file will be then automatically tagged with error description. Acknowledgements The research leading to these results has received funding from the European Community's Seventh Framework Programme (FP7/2007-2013) under grant agreement no. 218086. References [1] INDECT: For the Security of Citizens. Indect [online]. 2012. Available at: http://www.indect-project.eu/. [2] ZDRALEK, J., L. KAPICAK, A. FIGAJ, T. JANASIEWICZ, G. M. MOLINA and J. SAINZ. Data Formats and Protocols for Information Handling in INDECT Portal: Deliverable 6.3. In: Indect: For the Security of Citizens [online]. 2010. Available at: http://www.indectproject.eu/files/deliverables/public/deliverable-6.3/view. [3] VOZNAK, Miroslav. Voice over IP: College book. 1st. ed. Ostrava: VSB-Technical University of Ostrava, 2008. ISBN 978-80-248-1828-3. [4] VOZNAK, M., L. KAPICAK, J. ZDRALEK, P. NEVLUD and J. ROZHON. Video Files Recording and Playback with VoiceXML. In: 10th WSEAS International Conference on SIGNAL PROCESSING, ROBOTICS and AUTOMATION (ISPRA '11). Cambridge: WSEAS Press, 2011, pp. 350-354. ISBN 978-960-474-276-9. [5] Simena network emulator. Netscout [online]. 2012. Available at: http://www.simena.net. [6] MEGGELEN, J., J. SMITH and L. MADSEN. Asterisk: The Future of Telephony. 2nd ed. Sebastopol: O'Reilly Media, 2007. ISBN 978-05-965-1048-0. [7] HALONEN, T., J. ROMERO and J. MELERO. GSM, GPRS and EDGE Performance Evolution Towards 3G/UMTS. 2nd ed. West Sussex: John Wiley & Sons, Ltd, 2003. ISBN 9780470866948. [8] MSU Quality Measurement Tool: Metrics information. Everything about the data compression [online]. 2011. Available at: http://www.compression.ru/video/quality_measure/info_en.htm l. [9] XIAO, F. DCT-based Video Quality Evaluation: Final Project for EE392J. In: MSU Quality Measurement Tool: Metrics information [online]. 2000. Available at: http://wwwiso.compression.graphicon.ru/index_en.htm. [10] WATSON, A. B. Toward a perceptual video quality metric. In: Documents & Resources for Small Bussinesses & Professionals [online]. 1998 [cit. 2012-10-29]. Available at: http://www.docstoc.com/docs/80421243/Toward-a-perceptualvideo-quality-metric. [11] DOLNAK, I. and L. SCHWARTZ. Service Availability Issues in VoIP Networks. In: Digital technologies 2010: 7th International Workshop on Digital Technologies, Circuits, Systems and Signal Processing. Zilina: EDIS ZU, 2010. ISBN 978-80-554-0304-5. [12] DURCEKOVA, V. and L. SCHWARTZ. Network security attacks. In: Digital Technologies 2010: 7th International Workshop on Digital Technologies, Circuits, Systems and Signal Processing, Zilina: EDIS ZU, 2010. ISBN 978-80-5540304-5. About Authors Lukas KAPICAK received his M.S. degree in telecommunications from VSB-Technical University of Ostrava, Czech Republic in 2007. Since 2007 has been studying Ph.D. degree at the same university. His research is focused on wireless transmission and data flow analysis, simulation and optimization. Pavel NEVLUD received his M.Sc. degree in telecommunication engineering from VSB-Technical University of Ostrava, Czech Republic in 1995. Since this year he has been holding position as an assistant professor at the Department of Telecommunications, VSB-Technical University of Ostrava. The topics of his research interests are communication technologies, networking and security. Martin MIKULEC received the M.Sc. degree from VSB-Technical University of Ostrava, Faculty of Electrical Engineering and Computer Science in 2011. Currently, he is working toward the Ph.D. degree at the Department of Telecommunications, Faculty of Electrical Engineering and Computer Science. His research is focused on advanced services in IP telephony. Jaroslav ZDRALEK holds position as an associate professor with Department of Telecommunications, VSBTechnical University of Ostrava, Czech Republic. He received his M.Sc. degree in Computer Science from Slovak Technical University of Bratislava, Slovakia in 1977. He received his Ph.D. degree from VSBTechnical University of Ostrava in 2002, dissertation thesis „Diagnostic system without dismantling of locomotive controller“. His research is focused on fault tolerant system, communication technologies and coding.