Simplified Block Diagram 

Context in source publication

Context 1

... paradigm of research and development that enables multimedia content to move seamlessly between Internet and mobile wireless networks. With the benefit of increase in bandwidth in wireless networks, new access capabilities including mobile visual phones, personal digital assistants, and video streaming are pervading peo- ple’s everyday life. Such mobile web services provide us an enhanced ability to access to Internet content anytime, anywhere, and from any device. However, the converging technologies are the results of individual uncoordinated studies. Their design requirements and criteria are not related to each other’s. Hence, most of the time there is plenty of work to be done while connecting different kinds of networks. For example, Internet is based on TCP/IP transmission protocol, which provides reliable connection oriented transmission mechanisms. Applications utilizing TCP are not affected by the floating bandwidth capacity of the worldwide Internet network. On the other hand, multimedia content delivery is very much dependent on the capacity and the quality of the connection between two end points. Therefore, it is diffi- cult to carry multimedia data using TCP on the Internet [1]. There are a number of solutions that address this problem [2]. When wireless/mobile networks come into the picture, other problems arise due to the nature of communication in noisy channels [4]. Problems mainly fall into two catego- ries: coding techniques and transmission mechanisms. Some of them are, multi-path fading, shadowing, inter- symbol interference, and noise disturbance. There are many solutions to the defined problems of wireless/mobile networks [5, 6, 7]. On the other hand, the advances in computer architecture, display devices, power supplies and VLSI techniques opened a new dimension [8]. Computing devices are at pocket size now [9, 10]. These devices cannot be thought without Internet connectivity. Nowadays, there are some devices even with more than one network access, such as GSM and Wi-Fi or GSM, IR and Bluetooth. The multimedia capabilities of these devices are becoming comparable to the desktops. Web services of multimedia are within the reach of small devices now. Rapid growth of the mobile device market and wireless/mobile networks pushes us to think about delivering multimedia web services to the mobile world. With the idea of bringing multimedia web services to the mobile world, it is necessary to construct such a system that enables both fixed and mobile users to access content. It is our requirement to design a system so that content is available from any device anywhere. In this paper, a multimedia web service is proposed as a universal multimedia access system. The system is able to deliver live or stored video to mobile and fixed clients. Both wired and wireless networks can be used to access the video content. Although there are a number of similar systems in the market, our implementation differs from them in a number of points. First of all, a customizable and extendable architecture is required. Portability of object code has to main- tained since there will be different target clients that run operating environments of different vendors. Moreover, in order to measure the performance of the system, we should be able to access the inner parts of the implementation and have access to the source code. In the context of this paper, you will be reviewing an object based; multimedia web-service that is customizable, expandable and portable such that different targets running different operating environments can be clients. In this section, implemented system architecture is ex- plained. Deployment view of the architecture is illustrated in Figure 1. System consists of cameras up to 16. Server computer processes videos captured from the camera and recorded video is also streamed into client computers as well. Some of the client machines are desktop PCs and connected by Ethernet networks. Others are handheld devices (e.g. Compaq Ipaq 3870) and connected via Bluetooth connection. Compaq Ipaq PocketPC has a 200 Mhz processor. Server computer is a desktop PC and has a 2 GHz Pentium 4 processor. Bluetooth network has a bandwidth of 115 kbps, which is relatively small compared with 10 or 100 Mpbs Ethernet bandwidth. Bluetooth network has also more noise and bit errors due to its wireless connection. Due to bandwidth limitations, video data has to be encoded according to the limiting bandwidth of Bluetooth network. In order to avoid developing different client programs and then porting the code, we have selected JavaTM [11] as the development language. JavaTM (write once run anywhere) eliminates the porting of client code. The PocketPC has limited processing power and it is not sufficient to efficiently decode H263 coded video streams in real-time. Thus, a native, custom, efficient H263 decoder plug-in for JMF has been implemented and attached using Java Native Interface (JNI) [11]. As seen in Figure 2, video is encoded using H.263, streamed over Bluetooth using RTP, received and decoded by RTP and H.263 decoders and rendered in the client device. Several experiments are carried out to evaluate the performance of the system. In these experiments the variance of instantaneous frame rate and the difference between average instantaneous frame rate and source frame rate is calculated. The following parameters have been measured for each frame 1. Frame number : The order of frame in the stream. 2. Frame processing time : How long it takes to decode current frame. 3. Instantaneous frame rate : Reciprocal of frame processing time. The default parameters for all the experiments are as fol- lows: 1. Video sequence duration is 60 seconds. 2. Source frame size is 288x176. 3. Encoder output frame size is 128x96. 4. Color depth of the source is 24 bits (16,7 Million different colors.) In order to run the experiments under controlled input source, a recorded video has been used. The source rate has been adjusted to 5, 10, 15, 20 and 25 fps. The bit rate at the encoder output versus source frame rate is shown in Table 1. All the values obtained are less than the maximum bit rate available for the Bluetooth connection (115,200 bps). Thus, the network is not congested and we can assume that no packet loss shall occur due to network congestion. The following results shown in Figure 3 are the displayed frame rate with different source frame rates and intra frame rates. Experiment data is collected by playing the video for five times. The following observations are made on the obtained results: 1. The decoder can catch-up with a source rate of approximately 5 FPS. When the source rate is increased, the difference between sources and displayed rates increases. The decoder lags the source rate and more frames are dropped. 2. Changing intra frame rate with a constant source rate does not affect the displayed frame rate. In other words; from the decoder point of view it is not important whether the input frame is I frame or P frame. 3. Changing source rate with a constant intra frame rate reduces the performance of decoder. The de- coder cannot catch-up with the source when the source rate is increased. According to the observations, the source rate is selected as 5 FPS. Slower source rates could also be used. However, the quality of viewing experience degrades significantly. For the selected source rate, changing intra frame rate slightly changes the decoder performance. However, it seems, from the difference plots, that 15 intra frame rate performs slightly better. Thus, the intra frame rate is selected as 15. The aim of this work is to implement a universal multimedia system, which enables its users to access multimedia content over wireless networks, using different computing platforms. A number of different technologies have been involved in the study. Namely, Java technology is selected as the middleware for the software, Bluetooth networks connect the mobile devices to the network and PocketPC devices are the platform of choice when mobility is required. Also, several experiments have been conducted in order to evaluate the performance of the implemented decoder of mobile devices. Experiments have shown that even the small and limited (from the performance point of view) devices can be used as mobile stations to access multimedia content. Although, it is not possible to obtain outstanding results, current technology can satisfy the needs of someone with relax requirements. Particularly, the best decoder performance is obtained when the source frame rate is 5 FPS and one intra frame is inserted in a 15-frame group. It is estimated that with the increase in the computing power available to mobile devices, the decoding performance will not be a matter of concern in the near future. According to the results, wireless network performance is satisfying for a Bluetooth connection when there is a single mobile device. The resulting system is far from complete to have commercial value. Moreover, building a commercial system is out of the scope of this study. A real universal multimedia access system requires robust handling mechanisms, user- friendly interfaces, management functionality and embed- ded maintenance support. Hence, more work should be done in case the system is to be commercialized. On the other hand, the system provides a good base for research. It provides easy to use interfaces for developing experimental software. The following subjects are some of the research areas that can be experimented using the provided system. The mobile device can be used as the source of content, and video conferencing between two mobile devices can be evaluated. The system has not been tested under heavy network load. Other experiments can be done in order to evaluate the performance of the system, such as using higher bit rate video streams or simultaneous access of different devices. The system employs Bluetooth for ...