Author Guidelines for 8



Multimedia Applications and End Systems (Devices)  

Chao-kai Ching, George Lin and Yevgeniy Razuvayev

Department of Computer Science and Engineering, The Ohio State University

(ching.17, lin.539, razuvayev.1)@osu.edu

Abstract

With current advanced in network, multimedia, computer engineering technologies, in the future, it is highly possible that our desktop computers will not be the major end systems running multimedia applications. This paper focuses on three non-desktop end systems and discusses how they support multimedia through network in their own special environment, which provides different challenges for the device designers. Additionally, the hardware of each end system is introduced with discussion of how the problems for providing multimedia support in each end system is solved, or at least compromised.

1. Introduction

With current advanced in network, multimedia, computer engineering technologies, in the future, it is highly possible that our desktop computers will not be the major end systems running multimedia applications.

This paper focuses on three non-desktop end systems and discusses how they support multimedia through network in their own special environment, which provides different challenges for the device designers. Additionally, the hardware of each end system is introduced with discussion of how the problems for providing multimedia support in each end system is solved, or at least compromised.

The first category of devices introduced is about TV over the Internet, which enable users to watch TV anywhere. Following the TV over the Internet is the ZEUS-TS system, which allows surgeons to perform remote surgery. Finally, the technology and framework of telerobotics are discussed.

2. Entertainment

One of the most common and most profitable uses for multimedia is in the area of entertainment. Music, pictures, movies, and video games have been sources of revenue for many companies for years. With the growth of the Internet, more and more resources are being used to send multimedia to computers through the Internet. As the amount of bandwidth available to users steadily increased over the years, the amount of bandwidth being used for multimedia entertainment has also increased. With broadband Internet available in the home of many users, video files are becoming more common on the Internet. Videos can be downloaded or streamed by home viewers and some devices have been created to expand the available online content for the video watcher on the go.

2.1 TV Over the Internet

The ability to watch live TV over the Internet has been a previously unexplored idea. In recent years, several companies have developed devices to allow people to watch live TV programs over the Internet. These devices take a TV signal and compress the video to send over the Internet for a client to watch. The idea behind sending TV over the Internet is to allow viewers to watch live TV anywhere that there is a high-speed Internet connection. This could allow a frequent traveler to watch local TV while they are away from home so that they will never have to miss a sports game or their favorite TV show. TV over the Internet also allows for people to watch TV at Internet hot spots that can be found in many places around the world including cafes, malls, and even some McDonald’s restaurants. TV Over the Internet is also used to watch TV from far away locations such as other countries. A European soccer fan living in the United States of America can use TV Over the Internet to watch all of his favorite team’s games while being on the other side of the Atlantic Ocean. TV2Me, Slingbox, and Sony LocationFree are three hardware devices that act as live TV video servers that can send live TV video over the Internet. Orb is a software application and a service for Windows XP that allows a user to use his home PC to send live TV video over the Internet.

2.1.1 TV2Me

The TV2Me is one of the first devices created to allow for TV over the Internet. Its creator, Ken Schaffer, also invented the wireless microphone and the wireless guitar. Schaffer’s motive for creating the TV2Me was to be able to watch live Russian TV while he was in America and to watch live American TV while he was in Russia. While streaming video over the Internet is not new technology, the quality of streaming video tends to be lackluster. The impressive aspect of Schaffer’s invention is the quality of the video that is sent across the Internet. Robert Cringely states that “…at 384 kilobits-per-second using hardware encoding only on the sending end (the receiver is software-only) and he [Ken Schaffer] can watch perfectly viewable television that has run through 20+ hops from Moscow”[3].

[pic]

Fig. 1. TV2Me connection [1]

The TV2Me is an appliance that is connected to a TV signal and an Internet connection. The TV2Me can then send the video to a computer with an Internet connection and the necessary software. The video is sent over the Internet so the viewing computer can be located anywhere in the world as long as it is connected to the Internet. The TV2Me hardware is actually a Dell PC with a custom video capture card to preprocess and encode the video using an MPEG-4 encoder. The details of the preprocessing were not revealed but the preprocessing is credited for the superior quality of the TV2Me streaming video as the MPEG-4 encoding is used in other video streaming applications. MPEG-4 is an open international standard for multimedia and does not refer to a specific video codec.

Many multimedia codecs are MPEG-4 codecs and the following table lists some of those codecs as well as their applications. The newest MPEG-4 codec is AVC and it is designed for high quality video and is even used for Internet video. Apple Quicktime, a common application that is used by many Internet sites for video, is one of the applications that uses AVC.

[pic]

Fig. 2. MPEG-4 Codecs [4]

Despite the advancements in sending live TV video over the Internet made by TV2Me there are still several setbacks. TV2Me is only a point-to-point system where there can only be one viewer per TV2Me system. TV2Me also requires a 384 kpbs upstream bandwidth for the TV2Me system with a recommended 512 kpbs or 768 kpbs upstream bandwidth to allow for growth. Many broadband connections still do not have an upstream bandwidth of over 384 kpbs so it will still be inaccessible to many broadband users. There is also a buffering delay when the channel is changed in TV2Me, which can hinder the viewing experience, as channel surfing will take longer. Viewers also connect to the TV2Me by specifying the IP address so users without a static IP address may have to constantly lookup and remember the IP address for the TV2Me in order to connect to it. The final hurdle preventing TV2Me from becoming a common household appliance is its price tag of $4,750.00 for a pre-configured system with a potentially higher price tag for a custom designed system. The high quality streaming TV video of TV2Me comes at a high cost but Schaffer is hoping to reduce the cost of the TV2Me to under $1000 in the recent future but for the moment TV2Me is not a mass market offering. Other companies have also entered the TV over Internet market and offer their own solutions to allow for remote viewing of TV programs to the mainstream market.

2.1.2 Slingbox

The Slingbox is one of the competitors of TV2Me and offers a device that sends TV video over the Internet at a cost of $249.99. The Slingbox works in a similar manner to TV2Me but it encodes the video into Windows Media 9.0 format and uses a proprietary QoS software that optimizes the video based on the available bandwidth. This stream optimization is code named “Lebowski” and is designed to change the video parameters in real time by monitoring the network and making changes depending on the media, network, and the viewing system.

The Slingbox uses a technology called Slingstream™ to reduce delay when sending control signals since the Slingbox does not use a traditional buffering scheme. This means that changing channels will not result in as much of a delay since the video does not need to be buffered. The technology also analyzes the video and audio streams and sends it back to the encoder to alter the compression process depending on the type of video being viewed such as if it is a high motion sporting event or a low motion news cast.

[pic]

Fig. 3. Adaptive Compression [8]

Sling media, the creators of the Slingbox, also provide a service to find the IP address of the Slingbox system when away from home. This free service is called Finder and works by having servers remain in constant contact with the Slingbox and having users connect to their Slingbox through the Finder service. By using the Finder service the user does not have to remember the IP address of the Slingbox to connect to it.

The Slingbox is supposedly capable of working with a mere 50 kpbs bandwidth but to get a decent quality video stream it is recommended to have a transfer rate of 220 to 400 kpbs. The Slingbox can also only have one computer connect to it to protect the rights of copyright holders. The Slingbox attempts to bring live TV over the Internet to the mainstream user and it is available for purchase in some retail stores such as CompUSA or Bestbuy. It has features that make it more user-friendly than the TV2Me and is offered at a much lower price. However the video quality of the Slingbox is not as impressive as the TV2Me and Rik Farlie described the video quality of the Slingbox as, “… just a little better than the quality of streaming Web videos” [7].

2.1.3 Sony LocationFree

The first entry by a major consumer electronics company into the TV over Internet market is the Sony LocationFree. The Sony LocationFree is available in different packages and can be purchased as a base station alone or in a package where the base station is included with a LocationFree TV. The LocationFree TV is a LCD TV that can function as the receiver of the video stream and can also be used to surf the Internet. The LocationFree requires a bandwidth of at least 300 kpbs for good quality video. The base station alone costs $349.99 and has similar features to the Slingbox. Like the TV2Me, the LocationFree also uses a MPEG-4 encoding and it incorporates a DNS service integrated into its LocationFree Player Pak and LocationFree TV so the user doesn’t need to remember the IP address of the base station. The LocationFree can have four devices registered to use it and can work with a PC, Sony PSP, or the Sony LocationFree TV. Despite being able to work with up to four devices, the base station can still only send video to one device at a time. The LocationFree is also designed to work on both a home network or over the Internet and has superior quality to TV2Me when both are used within a home network. The TV2Me has superior video quality when both devices are used to send TV over the Internet.

2.1.4 Comparing the Devices

Video quality is a subjective evaluation that varies depending on the viewer. The typical way to determine which device has better video quality is to compare the videos from each devices side by side and determine which video looks better since there is no quantitative measurement for video quality. Of the three hardware devices designed for sending TV over the Internet, the TV2Me is considered to have the best video quality. The Slingbox and the LocationFree provide better than standard streaming video quality but do not quite match the video quality of TV2Me when using the same amount of bandwidth. They also come at a significantly lower price than the TV2Me. The table below provides a summary of the different devices that can send live TV over the Internet. The listed bandwidth is the minimum recommended bandwidth for good quality video and the quality description is in relation to other video streaming applications.

Table 1. TV Over the Internet Device Summary

2.1.5 Orb

The previously described devices are specific hardware solutions for sending TV over the Internet. Orb Networks provides a software option for TV enthusiasts to view TV over the Internet if the user’s home PC meets the hardware and software requirements to use Orb. To stream TV over the Internet, the user needs to have a Windows XP operating system and a compatible TV Tuner with a hardware MPEG-2 encoder. The user then downloads the Orb software and registers with Orb so that they can log in to the Orb site and connect to their home system running the Orb software from a remote system.

The remote system can just use a Web Browser and a video player such as Windows Media Player or Real Player to watch the TV video. Orb also allows the user to stream other digital media from their home PC such as photos, music, or videos. 300 kpbs can provide standard TV resolution using standard compression when using the Orb service. The best feature about the Orb software and service is that it is free if the user’s system already meets the requirements and it can also work compatible cell phones and PDAs. The Orb software enables users to use their home PC as a video server and allows the user to view their multimedia files using other devices besides just a computer.

3. Remote Surgery

Remote Surgery, also named as telesurgery, enables the doctor to perform surgery on a patient even when they are not physically in the same location [13]. The ability to perform telesurgery depends on the advance in both technologies of telecommunication and telerobotics.

3.1. Operation Lindbergh

The project of Operation Lindbergh, the first transatlantic remote surgery in 2001, serves as a perfect example to demonstrate the connection between telerobotics and telecommunication. Section three will cover telerobotics in more detail; this section will focus on the telecommunication, and only mention teleroboitcs part when needed.

3.2. The ZEUSTM System

ZeusTM is a robotic surgical system to perform endoscopic microsurgical procedures, a type of minimally-invasive surgery. Although the ZeusTM system itself can not be used automatically to perform remote surgery, its master-slaver configuration allows s the project designer to use it as a drop-in component without modifying the robotic system and preserve its fail-save features. The system is composed of two major parts, the master console where the surgeon performs the surgery and a set of robotic arms and camera-control equipment mounted on the operating room table [2]. Figure 1 shows the surgeon-site and Figure 2 depicts the patient-side equipment.

[pic]

Fig. 4. Surgeon –side equipment [13]

[pic]

Fig. 5. Patient-side equipment [13]

3.3. The Zeus-TS System

In order to support remote surgery, following equipments have been added. Two standard Pentium-based computers running VxWorks real-time O.S. with 100 base-T Ethernets are included on surgeon and patient side, and the patient and surgeon equipments are connected to the computers. Figure 3 indicates the modified version of the ZEUS system [13].

[pic]

Fig. 6. Telesurgical system after separation [13]

In the modified system, all the signals generate from the surgeon console and feedback signals from the patient side are communicated through the network connection. All the video codec is processed through different hardware and transferred through the network connection.

3.4. Network Communications

The surgeon-side and patient-side subsystems are connected through a dedicated 10 Mb/s CBR ATM circuit with OC-3 fiber. Standard UDP/IP protocol is used for network transmission.

3.5. Bandwidth Allocation

In order to achieve on time delivery of the high quality video signal in PAL format generated by the endoscopic camera, the technique of over provision is used. Seventy percent of bandwidth is reserved for video transmission. Other bandwidth is used by robot control signals transmission, which, including UDP/IP overheads and ATM/SONET framing as well as all embedded serial communications streams, are less than 60 Kb/s [13].

3.6. Package Format

Included in the data package is the information of robotic operation e.g. sensing and positioning the robots, and that of recovery from communication losses. Information used to maintain the fail-safe behavior of the Zeus system in case of communication breakdown is also added into the package. In order to detect errors and maintain order of the packages, each package has a package checksum and a sequence number. Both surgeon and patient side machine record the statistics of package arrival as an indicator of communications success and include the information within in the package sent [14].

For all robot-related data in the packet, the strict in-order delivery is not a concern. The only requirement for the packet is that it is not corrupt, i.e. passing the checksum, and is newer than the previous received one. Any robot-related data fulfill the above requirement are sufficient to provide robot position and feedback information for necessary robot operation. The advantage of sending all robot-related data in packets in an absolute form is that packages can be “dropped or even corrupted (hence discarded after error detection) due to communication bit errors without unduly affecting the in-progress operation as long as a certain minimum arrival rate of good packets is maintained [14].”

3.6.1. Surgeon-site package format. The Zeus system is designed in the master-slave fashion, and the surgeon-site serves as the master. Surgeon-site sends the data about machine console operation and receives feedback data from patient side. Figure 4 shows the robot control related data in surgeon-site package.

[pic]

Fig. 7. Surgeon site robot control data [14]

3.6.2. Patient-site package format. Comparing to the surgeon-site package, less data are sent from the patient-site to the surgeon-site. Figure 5 shows the robot control related data in patient-site package.

[pic]

Fig. 8. Patient-site robot control data [14]

Additional to robot control data, both surgeon and patient site package contains the following information (indicated in figure 6) to support and enhance communication reliability.

[pic]

Fig. 9. Other data for stable communication [14]

3.7. Communication anomalies and their solutions

3.7.1. Delayed Packets. For any real time application, timely delivery of process data is the most important consideration. In the remote surgery application, due to the high quality video requirement, the project designer decides to use over provision to achievement the latency requirement. The communications latency on the ability of a laparoscopic surgeon to perform various surgical manipulations should not exceed 700 ms [13]. The project has limit latency within 330 ms.

3.7.2. Single- or multiple-bit errors in a packet or Fragment. The Zeus-TS does not provide specially designed protocol to handle bit errors, but relies on the pre-built error checking facilities in the network. One of the reasons for lacking customized error detecting protocol is that the probability of a bit error in current high quality optical fiber links is less than 10-9. Therefore, at ATM level and UDP level, header checksum is used; at Ethernet level, CRC is used [13].

3.8. Dropped packets and out of order packets

Because the designer uses UDP/IP protocols, the ability of detecting dropped packets and out of order packets based on the sequence number is put upon the application layer running in the two telecommunication computers in the surgeon and patient side. Additionally, the system requires that each package received must be acknowledged [13].

4. NeuroMaster

NueroMaster (NM) is another telesurgical system for minimally invasive frameless neurosurgery. The major components of the system are similar to the Zeus-TS, which also includes the robot arm with its controller, the 3D vision system, the remote control system, an the network. Figure 10 depicts the structure of the system [15].

[pic]

Fig. 10. Structure of NeuroMaster system [15]

The major difference between NeuroMaster and Zeus-TS is the network component. For Zeus-TS, expensive, dedicated ATM is used to provide enough bandwidth for high quality video in PAL format. However, the use of the dedicated network is the major reason why the gall bladder surgery requires $1 million to perform comparing to a typical one costing only $2,000. On the other hand, the NeuroMaster can perform the remote surgery over ADSL and Modem.

4.1. The Video Codec in NeuroMaster

The main reason why NM is able to use ADSL instead of ATM depends on the preprocessing of the video image. Totally, there are three resources of the visual information, one video image for global vision, and two video images for local visions. In order for timely delivery of the three video images, all the three video images are first captured as white and black 8bit video BMP images. Depending on the sources of the visual information, the global one is sampled from PAL into CIF and two video images from local visual information are sampled from PAL into QCIF. Finally, all the BMP images are converted into JPEG format before [15].

4.2. Further investigation for the ADSL in NeuroMaster

However, the comparison made here requires further investigations. Even though both of the system is designed for minimal invasive surgery, the two remote surgeries performed are totally different, one for the gall bladder surgery, and one for cerebral-hemorrhaged disease. The different types of the surgery require different medical data to be processed during the communication, which lead to the difference complex in network and multimedia support. For instance, the video quality requirement in NeuroMaster system is lower than the Zeus-TS. Second, Zeus-TS performs a transatlantic remote surgery between France and USA, but NeuroMaster performs the surgery between Beijing and Shenyang instead. The geological difference affects the choice of the network considerably. Therefore, whether NeuroMaster has developed a more sophistic scheme in terms of network and multimedia support remains debatable.

5. Robotics Environment and Telerobotics

Robotic Environment and Telerobotics is one of the growing and expanding fields of research today. Today we can see robots being employed on the jobs and places where it is deemed too dangerous for humans to venture to. Army uses remote robots to fly spy planes missions over hostile territories of Iraq and Afghanistan. Police uses remote robots to resolve hostile situations such as exploding device disarmament and subduing a suspect in barricaded situation. Even NASA uses remote robots to explore the distance planets in our solar systems. In this section, the general architecture and requirements of remote robotics will be discussed and presented [16].

[pic]

Fig. 10.

5.1. ATM Network Architecture

Remote robotics is the primary example of ATM (Asynchronous Transfer Mode) Network, and uses end-to-end communication between the master side and the slave side. This communication must be high speed and robust to accommodate the requirements established by remote robotics systems.

The master side is the end-point where human operator is stationed and operates the slave side, or the robotic side.

The slave side is the end-point where robotic system is present and continuously sends the state information back to the master side.

5.2. Architecture of Telerobotics

[pic]

Fig. 11. [17]

The architecture of telerobotics depends on the complexity of remote robot system and its tasks. However, almost all remote robots system architectures share the following common characteristics.

The architecture of the master side is relatively simple and consists of a human operator and a terminal display or control device. In most cases, the terminal and control device would be a software application, which runs on the operator’s workstation. Using the terminal and control device, operator is able to send commands to the robot. The terminal is then used to receive and view responses that robot sends back. The terminal usually displays the information about the current state of the robot, as well as any video feed received from a robot including the audio feed, if one is present.

The architecture of the slave side, or the robot side, consists of robotic system, which is the robot itself. They can be anything from a single robotic arm that picks up objects and moves it around, to complex robotic systems that move around. The major component of any robotic is the sensor unit that gathers all the information regarding the state of the robot as well as the environmental variables and information of the environment in which the robot is currently present. This information is then transmitted back to the operator’s terminal. In addition, this information can include the video and audio feeds, which allows the operator for better assessment of the current state of the remote robotic system and its environment.

5.3. Requirements of Telerobotics

In the previous section, three major components of telerobotics’ architecture were identified, the sensor, the terminal, and the video feed. All three of these components play the vital role in the telerobotics architectures and all of them have unique requirements that must be satisfied to ensure a functional operation of remote robotic system [17].

5.3.1. Sensor and Operator Terminal Requirements

The sensor is one of the major units of remote robotic unit. It gathers and stores information about the current state of the robotic unit, until that information can be relayed back to the operator.

While the idea of rampaging robots can be quite amusing, we all know how it is frustrating when even the simplest of devices refuses to behave as expected. It is for this simple reason that sensor must be accurate and redundant in its collection of information. One of the major characteristic of the sensor unit is that it must have a high degree of reliability. When sensor transmits the information back to the terminal, it must insure that no data loss occurred. If data lost detected, it must be minimized as much as possible. The reason for this is to minimize and eliminate the operator error due to outdated and inaccurate information being sent from the remote robotic system. The operator must trust the information he/she is provided to operate the robotic unit at is peak efficiency and safety.

Second major characteristic of the sensor unit is to provide accurate and high precision information. However, it is an inheriting flaw in all binary computers systems that a simple decimal number of “0.2” ([pic]) cannot be represented as a finite number in binary number system. Therefore, computer systems must do their best to approximate such numbers. The accuracy in remote robotics units is important because it can mean a fine line between success and failure if robotic unit must navigate and complete fine precision tasks such as high-explosives disarmament or complex navigation.

The final major characteristic of the sensor unit is to provide minimum end-to-end delay of all information it sends out to the terminal and the operator, and vice versa. It is crucial in most environments, to which remote robotic units are deployed, that operator should control the remote robotic unit in real time, and that means to minimize end-to-end delay as possible. Back to the example of explosives disarmament, the time is crucial and so the operator must be able to send commands to the remote robotic system without experiencing any soft of delay. The same condition applies to remote robotic spy and reconnaissance planes where accuracy and instantaneous delivery of data is of major concern [17].

5.3.2. Video and Audio Feed Requirements

The second major units of remote robotics systems are video and audio feeds that robots send to the operator’s terminal.

Video and audio are gathered by the remote robotic system and send to the operator terminal to allow operator for a better understanding of the environment the remote robotic system is located. In most cases, the video and audio is optional, however they provide an expanded and crucial information that sensor unit can not, and they also serve as a back-up alternative if sensor unit fails to replay accurate information back to the operator terminal.

As with sensor unit, video and audio feed must try to minimize the loss of video and audio data. While video and audio data is of less importance then sensor data, it is still crucial to provide an operator with a clear and understandable video and audio, since operator maybe relaying on it as well as the sensor data to operate a remote sensor unit.

In addition, end-to-end delay must also be at a minimum to provide a smooth video and audio, and to minimize jitters. However, unlike the sensor data, the quality and rate of video and audio data can be adjusted to allocate for delay in the network. Thus, allowing video and audio feed can be more flexible [17].

5.4. Telerobotics Network

As we discussed the architecture and the requirements of telerobotics and remote robotic systems, the last question remains of what kind of network environment is needed to support such end systems.

For sensor unit communication, where the emphasis is on reliability and efficiency, the TCP protocol would be sufficient to support all communication between sensor unit and operator terminal. The TCP protocol would guarantee a fair transfer of sensor data and in-order arrival with minimum of delay and loss, which will fulfill our requirements for reliability and efficiency.

For video and audio feed transfer, where the emphasis is on the delivery, the UDP protocol would be sufficient. The UDP protocol applies best effort, and since both video and audio data should be transmitted at best effort and allowed to have some jitters, it is sufficient to use simpler protocol, such as UDP, to transmit video and audio feed [16].

References

[1] “TV2Me”, .

[2] Seth Schiesel, “I Want My Moscow TV”, The New York Times, , December 2, 2004.

[3] Robert Cringely, “Come to Daddy: How Ken Schaffer’s TV2ME (or Something Just Like It, But Cheaper) Will Change Television Forever”, I, Cringey, PBS, , October 28, 2004.

[4] ”MPEG Industry Forum”,

[5] “Sling Media”, .

[6] Rik Farlie, “Slingbox Puts TV Content on Your Notebook”, Computer Shoppe, CNET, , January 6, 2005.

[7] “CES 2005--The Next Big Thing: Video-on-the-Go ("Vidi-Go")” , Broadband Home Report, , January 24, 2005.

[8] Robert Flood, “Using IPTV to Redirect Television Viewing –Place Shifting” , IP Television Magazine, , September 2005.

[9] “Sony LocationFree”, .

[10] Seth Schiesel, “Head to Head; Television From a Distance: Comparing Two Approaches in the Field”, The New York Times, , December 2, 2004.

[11] “Orb Networks”,

.

[12] J.D. Lasica, “The Engadget Interview: Jim Behrens, CEO of Orb Networks”, Engadget,

, June 20, 2005.

[13] S. E. Butner and M. Ghodoussi, “Transforming a Surgical Robot for Humna Telesurgery,” IEEE Transactions on Robotics and Automation, vol. 19, No. 5, pp 818-824. Oct. 2005.

[14] S. E. Butner and M. Ghodoussi, “A real-time system for telesurgery,” in Proc. IEEE Int. Conf. Distributed Systems, Mar. 2001, pp. 236-243

[15] C. Meng, T. Wang, W. Chou, S. Luan, Y. Zhang, and Z. Tian, “Remote Surgery Case: Robot-Assited Teleneurosurgery,” in Proc. IEEE Int. Conf. on Robotics & Automation, Apr. 2004, pp. 819-823.

[16] Ahmad T. Al-Hammouri, Vincenzo Liberatore, Huthaifa A. Al-Omari, Stephen M. Phillips, “Transversal Issues in Real-Time Sense-and-Respond Systems,” Case Western Reserve University, Arizona State University, 2002;

[17] Klara Nahrstedt, Jonathan Smith, “QoS Negotiation in a Robotics Environment,” Distributed System Laboratory, Computer Science Department, University of Pennsylvania;

cis.upenn.edu/~dsl/read_reports/Negotiation.ps.Z

[18] Klara Nahrstedt, Jonathan M. Smith, “An Application-Driven Approach to Networked Multimedia Systems,” Distributed Systems Lab., University of Pennsylvania, Philadelphia, PA;

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download