Enhancement of wireless LAN for multimedia home ... - IEEE Xplore

80

IEEE Transactions on Consumer Electronics, Vol. 51, No. 1, FEBRUARY 2005

Enhancement of Wireless LAN for Multimedia Home Networking Klaus Jostschulte, Rüdiger Kays, Member, IEEE, and Wolfgang Endemann

Abstract — This paper describes the problems of wireless LAN usage in the user’s home with focus on high quality, realtime media transmission. Results of physical channel measurements with respect to fading effects and channel power are analyzed and experimentally obtained link data rates are presented. Methods for a transmitter based link adaptation are described and evaluated to increase the reliability of the physical data transmission1. Index Terms — Wireless LAN, home networks, AV network, channel models, fading, diversity, link adaptation, antenna diversity, dynamic frequency selection, IEEE 802.11, OFDM.

I.

INTRODUCTION

THE IEEE 802.11 standard family is a good basis for building wireless A/V-networks in the user’s home environment. With growing success of this technology mutual interference between nodes of the same household but also between nodes of adjacent networks will arise. In a network with off-the-shelf WLAN systems it is very likely that data rates and short latencies required for high quality, real time video transmission cannot be provided in such situations. Furthermore the quality of each individual link is influenced by the characteristic of the transmission medium. As an implication a constant data rate and a reliable quality can not be guaranteed over longer periods of time. In this paper we present results of our investigations towards enhancements for home AV-networks based on IEEE 802.11a/g. In section II the limitations of current WLAN technology are described and improvements with respect to the medium access protocol are briefly described. Section III gives the results of measurements in typical home scenarios. These results are both related to channel characteristics as well as achievable data rates for standard WLAN equipment. In section IV we give an overview on different concepts for linkadaptation of the OFDM-based physical layers. The potential of these methods is analyzed as well.

1

Rüdiger Kays is the head of the Communication Technology Institute, Department of Electrical Engineering and Information Technology, University of Dortmund, Germany (e-mail: [email protected]). Klaus Jostschulte and Wolfgang Endemann are senior scientists at the Communication Technology Institute, University of Dortmund (e-mail: [email protected] / [email protected]). Manuscript received January 7, 2005

Fig. 1. Scenario for wireless AV home networks

II. MULTIMEDIA HOME NETWORKS WITH WIRELESS LAN IEEE 802.11 is a widespread used technology for connecting computers in offices, public places, and also in the user’s home. The strong penetration of the market and also the low prices of the transceiver chipsets make it a good aspirant for further applications in the user’s home. One example is the distribution of audiovisual data in the entire house as shown in Fig. 1. At the moment the transmission of media data over IEEE 802.11 links is not as reliable as required for a broad consumer market. Of course both the standards as well as the equipment are in permanent refinement. But from the technical point of view limitations still exist with respect to the medium access (MAC) protocol and data transmission capabilities due to limitations of the physical layer. A. Limitations of the Standard MAC Protocol The MAC in IEEE 802.11 uses a CSMA/CA protocol. The collision avoidance is realized by means of a constant waiting period and an additional random waiting period between successive data packets. The resulting best-effort like transmission is not well suited for transmitting constant rate A/V-data. Any occurring file transfer from another station will potentially interfere with running media transmissions. The resulting breakdowns of the running program will not be tolerated by the user. One solution for this problem are modifications of the standard MAC protocol. The working group IEEE 802.11e is currently developing a standard extension that is targeting to

0098 3063/05/$20.00 © 2005 IEEE

K. Jostschulte et al.: Enhancement of Wireless LAN for Multimedia Home Networking

81

provide Quality of Service (QoS) on the wireless link. The basic element of this extension is the introduction of waiting periods with ascending duration for different traffic categories. However an unmodified employment of this protocol might still exhibit some problems in areas with a high number of nodes in a small area. So in former work we defined certain policies to rule the use of different constant waiting times for AV-data only within the range of one network. Any further random waiting time should be avoided to increase the efficiency of the protocol. Simulations of this protocol have been presented in [1]. B. Data Transmission Limitations In many cases data rate fluctuations of an individual link as well as link breakdowns might occur. These effects very often originate from the transmission channel properties. So knowledge about the channel’s characteristics in the house is necessary for the development of data transmission enhancements. For an acquisition of these characteristics we performed a large amount of transmission channel measurements as well as an analysis of the throughput behavior of standard equipment. The results of these investigations are presented in detail in the next section. III. LINK MEASUREMENTS

attic RF source

Fig. 2. The location of these measurements is a 3 story apartment in a multifamily residence. The area for NLOS-measurements is depicted.

In our measurements this scenario is realized with one device on the attic and the other device somewhere in the floors below, as depicted in Fig. 2. A. Channel Characteristics The measurements with respect to channel characteristics were designed to gain information like fading effects, channel power and spatial correlation. The set-up is depicted in Fig. 3. The radio transmitter consists of a signal generator emitting band-limited white noise with a bandwidth of 100 MHz modulated with a desired carrier frequency. Because all physical layers of wireless LAN had to be covered, carrier frequencies of 2.45 GHz and 5.775 GHz are used to allow assessments of the ISM bands at 2.4 GHz and 5 GHz. To ensure a good coverage the test signal is amplified to an output power of 40 dBm at 2.4 GHz and 37 dBm at 5 GHz. At the receiver side the signal is detected using a spectrum analyzer that is controlled by a PC. So an automatic data acquisition could be realized. To overcome the problem of strong spatial dependencies of the radio field the antenna is automatically positioned using a specially designed antenna positioner. With this device the antenna can be placed automatically within an area of 80 cm x 80 cm with an accuracy of 0.025 mm. The superstructural parts of this device are mainly build of nun-conductive material to minimize interference with the radio field. This positioner is controlled by the PC as well, so that noise generator 50 MHz

868 MHz RS 232

RS 232 IEEE 1284

PA

signal generator und modulator 300 kHz - 6,4 GHz

IEEE 802.3 control data

2.4 GHz

control acknowledge

control acknowledge

Detailed and reliable knowledge about the characteristics of the transmission channel is necessary for the improvement of data transmission. The objective is to enable a reliable data transmission over long periods of time. Furthermore the channel parameters are necessary for a realistic simulation of the overall network behavior. Recent literature (see e.g.[2],[3]) as well as the highly respected overview in [4] mainly describes channel measurements in office environments. Comprehensive information for the home environment is rare. Thus we performed measurements of the transmission channel as well as the resulting data rates. The location for these measurements is a multifamily residential building. The apartment chosen has three stories. The lower two stories are the living space, the highest story is an attic. The separation between the lowest level and the medium level is a concrete ceiling with an open stairway. The separation between the attic and the living space is a lightweight construction. Fig. 2 gives an overview of the test environment. The analysis comprises two typical types of transmission scenarios. The first one is a "line-of-sight" transmission (LOS). In this case both stations are usually in the same room with no obstacles in the direct line between the two antennas. The second scenario is a "non-line-of-sight" transmission (NLOS). This is the typical situation if two devices are placed in different rooms, so that the radio waves have to travel through doors, walls, ceilings, etc.

NLOS measurement area

5 GHz antenna positioner 80 x 80 cm 0.025 mm accuracy

spectrum analyzer 20 Hz - 7 GHz

Fig. 3. Measurement set-up for registering the channel characteristics.

82


0.9

LOS

0.8 0.7 0.6

NLOS

0.5 0.4 0.3 0.2

2.4 GHz

0.1 0

Probability P(fading ≤ x-value)

1

Fig. 4. CDF for maximum fading depths at 2.4 GHz in a 20 MHz channel.

positioning and data acquisition is done automatically. With the help of this arrangement parts of a room can be surveyed without any user interaction. An optional control connection also exists between the PC and the sender, so that parameters like output power, frequency range, etc. can also be controlled automatically. For the test home we have several thousand measurements available, forming a dense mesh of measurement points. One objective of these measurements is to quantify emerging fading depths within the WLAN frequency range of 20 MHz. The resolution bandwidth of the spectrum analyzer is chosen to 300 kHz. This is in the same order as the subcarrier bandwidth of the OFDM-based physical layers in IEEE 802.11a/g. So this analysis reveals the attenuation affecting one carrier in worst case. In Fig. 4 the cumulative density function (CDF) for the maximum possible relative fading within a 20 MHz frequency band is depicted. The term relative fading attenuation specifies the ratio between the highest and the lowest reading within 20 MHz. The figure shows that 50% of the positions have fading depths of more than 10 dB in the LOS case. For NLOS transmissions fading is even worse. Here the 50% value is at 12.5 dB. Another interesting result of these measurements is the remarkably high probability for fading attenuations of more than 30 dB in both scenarios. A similar analysis can be found, if the fading depth is analyzed for the channel with the highest and lowest reception power. This is relevant for systems using a dynamic frequency selection (DFS). The resulting CDFs for 2.4 GHz and 5 GHz are depicted in Fig. 5 and Fig. 6 respectively. These figures emphasize, that the choice of the channel with the highest average reception power usually delivers lower fading effects than the choice of the worst channel, as expected. The 50% value is 4.12 dB and 3.3 dB for 2.4 GHz and 5 GHz respectively.

best channel

0.9 0.8 0.7

worst channel

0.6 0.5 0.4 0.3 0.2

2.4 GHz

0.1

5 10 15 20 25 30 35 40 maximum relative fading attenuation (dB)

0

5 10 15 20 25 30 maximum relative fading attenuation (dB)

Fig. 5. CDF for maximum relative fading at 2.4 GHz for the IEEE 802.11 channels with highest and lowest channel power.

1 Probability P(fading ≤ x-value)

Probability P(fading ≤ x-value)

1

best channel

0.9 0.8 0.7

worst channel

0.6 0.5 0.4 0.3 0.2

5.8 GHz

0.1 0

2 4 6 8 10 12 14 16 18 maximum relative fading attenuation (dB)

Fig. 6. CDF for maximum relative fading at 5 GHz for the IEEE 802.11 channels with highest and lowest channel power.

Another interesting figure is the attenuation a signal undergoes while crossing one or multiple floors. The attenuation added for crossing one ceiling (lightweight construction) is 9 dB and additional 20 dB for crossing a 2nd ceiling (concrete ceiling with an open stairway). Furthermore the power distribution in a room is non-uniform. In the NLOS case there is an averaged 14 dB higher reception power in the middle of the room compared to near-wall positions. B. Data Rates For any application usually not the channel characteristic but the resulting data rate is important. So the same environment is used to carry out measurements of the resulting data rate of standard wireless LAN equipment. In general the theoretical data rate of wireless LAN is dependent on the physical layer and the MAC protocol. In [1] it is shown that the theoretical net data rate NR (neglecting transmission


83

54 Mbit/s

35 30 net rate (Mbit/s)

36 Mbit/s 25 24 Mbit/s

20 15 10 5

500 1000 1500 2000 Number of bytes per packet [NBP]

Fig. 8. Measured mean data rates (upper part) and minimum data rates (lower part) for different positions with an IP packet size of 1500 bytes.

Fig. 7. Theoretical data rates for a link between two WLAN stations with an OFDM based physical layer at different transfer modes.

errors) of a link can be calculated from the physical layer gross rate DR in terms of an efficiency coefficient η : NR = η ⋅ DR

(1)

According to [1] the coefficient η can be calculated using

η=

NBP ⋅ (1 − pc )

  ( NBP + 246 )   DR ⋅    ⋅ 4 µs + (177,5 µs )    DR ⋅ 4 µs  

.

(2)

NBP is the number of user data bytes per WLAN packet and

pc the probability for a collision on the transmission channel. The resulting data rates as a function of the packet size are depicted in Fig. 7 for different transfer modes. For a comparison of theoretical and practical data rates in home applications measurements of the resulting data rates at different positions are carried out. Again LOS and NLOS scenarios are under consideration. The NLOS values are generated with one device on the attic and the other device either at the upper (passing one ceiling) or the lower part of the living space (passing two ceilings). The measurements are done using a specially designed software, that transmits the test packets using the UDP protocol. Simultaneously a TCP connection over a physical link at 1 Mbit/s is established for control purposes. An arbitrary packet size can be selected. In our measurements IP-packet sizes of 78, 528 1028 and 1500 bytes are used. The results of the mean data rate of all measurement positions for a packet size of 1500 bytes is depicted in the upper part of Fig. 8. One result is that in a LOS scenario usually the data rate is near the theoretical value. This favorable situation no longer holds for a NLOS transmission. Passing the lightweight ceiling the mean data

Probability P(data rate ≤ x-value)

0 0

1 0.9 0.8

54 Mbit/s

0.7

1 Mbit/s

0.6 0.5 0.4 0.3 0.2 0.1 0 0,1

36 Mbit/s 24 Mbit/s 12 Mbit/s

11 Mbit/s 6 Mbit/s 1 data rate (Mbit/s)

10

Fig. 9. CDF for measured data rates at different physical layer modes for NLOS data transmissions across 2 stories.

rate is slightly decreasing for 54 Mbit/s and 48 Mbit/s due to frequent packet losses. Passing two stories the mean values for the high data rate modes is decreasing dramatically. Even the 24 Mbit/s transmission mode is no longer working properly. A simple rating of the mean data rate value does not give any information about an individual link. The problem is that wireless LAN links usually do not degrade gracefully, but abrupt. To emphasize this assertion the lower part of Fig. 8 displays the minimum data rate of all measurement points in the LOS and NLOS scenarios. Again it is obvious that in a LOS transmission with short distance even the minimum data rate is near the theoretical value. But even if the signal has to cross only one lightweight ceiling, links with transfer modes of 48 Mbit/s and 54 Mbit/s might vanish completely in worst case situations. And even the lower transfer modes do not all have the full performance because of very excessive packet losses. In case of crossing two ceilings, the minimum data rate vanishes completely for all modes but 1 Mbit/s and 2 Mbit/s. A more detailed analysis of the link behavior can be done using Fig. 9. Here the CDF of the measured data rates are

depicted for the NLOS transmission across two stories. It becomes obvious that a transmission with 54 Mbit/s (64-QAM with code rate R=3/4) is not possible in 75% of all measurement points. Only in less than 10% of all positions this mode offers data rates of more than 10 Mbit/s. But of course, the high discrepancy between the theoretical value and the measured values shows the high loss rate caused by the space varying character of the link. Another interesting curve is the one for 24 Mbit/s (QPSK with R=3/4). For this transfer mode only 18% of all positions do not have a link at all. In 60% of all positions a data rate of more than 10 Mbit/s is obtained. Furthermore it can be stated that only the mode with 1 Mbit/s is reliable for all positions. It offers data rates of more than 600 kbit/s in either case. One disillusioning interpretation of this result is that with current wireless LAN equipment a reliable transmission with high data rates can not be guaranteed for NLOS situations. So an unmodified deployment of wireless LAN for media transmission can not be recommended. IV.

LINK ADAPTATION

The measurements of the physical transmission channel and the link data rates reveal that an adaptation of the physical link is necessary to overcome the apparent problems. For OFDM based transmissions a variety of transmitter based link adaptation principles is known from the literature [5]-[8]. The results of our channel measurements give advice and show the potential for applying the following methods. A. Dynamic Frequency Selection (DFS) The coherence bandwidth

with

1 , 5σ t

σ t2 being

1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

LOS NLOS

2.4 GHz 5 10 15 20 frequency selection gain (dB)

25

Fig. 10. CDF for the achievable channel power gain using the optimal transmission channel at 2.4 GHz.

1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

BC of a transmission channel can

be estimated using

BC ≈

Probability P(gain ≤ x-value)



84

NLOS LOS 5.8 GHz 5 10 15 20 25 frequency selection gain (dB)

30

Fig. 11. CDF for the achievable channel power gain using the optimal transmission channel at 5 GHz.

(3)

the delay spread of the channels impulse

response [9]. So for a typical home environments with maximum delays of some ten nanoseconds and RMS delay spread values in the same range, the WLAN bandwidth and the coherence bandwidth of the channel are of the same order of magnitude. For this reason and as the measurements in Fig. 5 and Fig. 6 have proven, there is a good chance to find a transmission frequency that has a high average reception power and relative fading effects of less than 10 dB in more than 90% of all cases. So if some carriers of the OFDM multiplex are affected by fading or if the overall reception power is comparably low the whole transmission can be shifted to another physical channel. In Europe there is the choice for 13 partly overlapping channels with a bandwidth of 20 MHz in the 2.4 GHz ISM band and 19 non-overlapping channels in the range between 5 GHz and 6 GHz.

Taking into account the channel power of the transmission we investigated the effect of changing the transmission channel to a less disturbed frequency. The effect of this method on the channel power is depicted in Fig. 10 for 2.4 GHz and in Fig. 11 for 5 GHz. These figures present the difference between the power of the channel with the highest and lowest channel power. For 2.4 GHz this difference is up to 20 dB. So the choice of the correct channel might give a gain in overall reception power of 20 dB compared to the channel with the lowest reception power. In 50% of all cases this gain is at least 5.4 dB for a LOS transmission and 7.23 dB for a NLOS transmission. The fact that the gain is typically higher at NLOS situations is very interesting, because this is usually the more critical case for a link. At 5 GHz the situation is slightly different. Here there is usually a higher gain for LOS situations. But in either case the probability for a certain gain is usually higher compared to 2.4 GHz. So the 50% values are at 10.38 dB for a LOS situation and 9.45 dB for NLOS transmissions.

0

5

10 15 diversity gain (dB)

5.8 GHz 0.5 0

20

Fig. 12. CDF for the achievable diversity gain using two antennas with a distance of 5 cm (2.4 GHz).

B. Antenna Diversity Apart from the optimization due to channel selection antenna diversity is another very powerful method for link adaptation. Even a very simple antenna diversity with antenna switching can offer high gains. The reason for this effect is the spatial coherence of the channel. In [2] it is stated that for a channel situation with equally distributed angles of incidence and equal path amplitudes a spatial decorrelation can be assumed for distances

s > 0.4λ ,

0.6

(4)

with λ the wavelength of the radio signal. Taking this into account, a two antenna diversity with an antenna distance of 40% of the signal's wavelength would be sufficient. So even for 2.4 GHz this results in a minimum required antenna distance of 5 cm, for 5 GHz even less. Of course in reality the wave incidence is not as in the idealized case. The distributions of angles of incidence and path amplitudes will be different. For this reason we analyzed the spatial correlation of the radio fields in indoor environments. The result is that the correlation distance is still of the same order of magnitude of the wavelength. Similarly to frequency selection the possible gain for antenna diversity with respect to channel power can be analyzed. The CDF for the achievable gain, if two antennas with a distance of 5 cm are used, is depicted in Fig. 12 for 2.4 GHz and in Fig. 13 for 5 GHz. The reason, that both figures start at a value of 0.5, is due to the fact, that a second antenna is advantageous only in every second case. In both frequency areas there is a probability of about 10% that the gain is 5 dB or more. In some cases gains of nearly 20 dB with an antenna distance of 5 cm have been found. C. Subcarrier Equalization at the Transmitter A more sophisticated method for adapting the transmission to the characteristics of the medium is to pre-equalize the spectrum of the transmitted signal to the frequency response of

5 10 diversity gain (dB)

15

Fig. 13. CDF for the achievable diversity gain using two antennas with a distance of 5 cm (5 GHz).

the channel. The principle of this technology is depicted in Fig. 14. This technology makes use of the fact, that in case of OFDM based signal generation the frequency components can be adjusted very easily. If the profile of the transmission channel is known at the transmitter, the inverse channel frequency response can be applied before passing the symbols to the IFFT, so that each frequency component is adjusted with respect to the channel's transfer function. For wireless LAN there exist several ways to provide the transmitter with the channel's transfer function. The easiest method is to use the reciprocity of the channel, so that the channel profile of a received ACK-frame is identical to the one for transmissions in the other direction. Another way is to use control packets for transferring the desired adjustment profile to the transmitter. One constraint of link adaptation in wireless LAN that has an impact on this method is the need to limit the overall output power of the transmitter to the permitted value. Due to this a high gain for one strongly faded subcarrier involves a reduction of power for all other subcarriers. We performed simulations on the behavior of this type of link adaptations with respect to the resulting bit error rate. The results are depicted in Fig. 15 for maximum fading depth of 10 dB and 25 dB. These simulations show, that an inversion of the channel's profile results in an improvement compared to an unprocessed transmission. This even holds, if the fading depth is very high, so that much transmission power is spent on only a few strongly faded subcarriers. fading distortions

transmitter

receiver

channel

Estimation

2.4 GHz 0.5

0.7

FFT

0.6

0.8

MUX

0.7

NLOS

MUX

0.8

0.9

IFFT

NLOS

LOS

Equalization

0.9

1

DEMUX

LOS


1

85

Modulation



Fig. 14. Subcarrier equalization at the transmitter for link adaptation.

bit error rate

86


10

-1

10

-2

10

-3

10

-4

10

-5

10

-6

10

-7

10

-8 -9

10 5

unprocessed transmission channel inversion

adaptation as well as MAC parameter adaptation require methods for self organization of the network nodes. The development and implementation of these methods will be the task of future work.

Channel with 25 dB relative fading

ACKNOWLEDGMENT The authors would like to thank the "Deutsche Forschungsgemeinschaft" (DFG) for financially supporting this work.

Channel with 10 dB relative fading

REFERENCES [1]

10

15

20

25

30

35

Eb/N0 Fig. 15. Bit error rate behavior of a transmitter based subcarrier equalization at channel profiles with a fading depths of 10 dB and 25 dB.

D. Subcarrier Modulation Adaptation Changing the modulation on a subcarrier basis also offers further potential for improving the link quality, but is not compliant to current standards and is subject to further investigations. E. Transmit Power Adaptation For short distances between two devices less power than allowed is required to achieve a reliable transmission. As a result the overall spectral efficiency of a dense wireless LAN network can be increased. All presented methods for link adaptation show high potential for network improvements. Via in-band communication – probably using ACK-messages – but also via external networks the nodes may exchange information about the channel state. This information should be used to select an adequate channel and perform a carrier based pre-equalization. Furthermore antenna diversity will help to improve the performance. V. CONCLUSION IEEE 802.11 and its extensions are good candidates for media transmissions in the home. Problems exist with respect to the MAC protocol and the reliability of data transmission due to the channel characteristics. Measurements of the obtained data rate using current standard equipment also show the unreliable behavior of a standard wireless LAN link. The channel’s influence on the transmission can be reduced using methods of link adaptation. A high SNR gain is possible by means of dynamic frequency selection, antenna diversity, and pre-equalization. The potential of these methods has been derived using the results of the channel measurements. The link adaptation approach will increase the individual link data rate as well as the overall throughput. So a reliable distribution of A/V data can be possible in the home environment. In the consumer market administration by the user should be avoided as much as possible. Therefore the efficient use of link

[2] [3] [4] [5] [6] [7] [8] [9]

R. Kays, K. Jostschulte, W. Endemann, "Wireless ad-hoc networks with high node density for home AV transmission," IEEE Trans. on Consumer Electronics, vol. 50, May 2004, pp. 463-471 L.B. Bertoni, Radio Propagation for Modern Wireless Systems; Prentice Hall; 2000 J. Medbo, H. Andersson, P. Schramm, H. Asplund, J.-E. Berg, "Channel models for HIPERLAN/2 in different indoor scenarios," Tech. Rep. COST 259 TD(98)70, EURO-COST, Bradford, UK, 1998. H. Hashemi: "The Indoor Radio Propagation Channel," Proceedings of the IEEE, Vol. 81, No. 7, July 1993 M. Lampe, H. Rohling, W. Zirwas, “Misunderstandings about link adaptation for frequency selective fading channels,” PIMRC 2002, Sept. 2002 S. Pfletschinger, G. Münz, J. Speidel, “Efficient subcarrier allocation for multiple access in OFDM systems,” 7th International OFDM-Workshop 2002 (InOWo'02), Hamburg, Germany, September 2002 J. Jang, K. B. Lee, “Transmit power adaptation for multiuser OFDM systems,” IEEE Journal on Selected Areas in Communications, vol. 21, no. 2; Feb. 2003 T. Keller, L. Hanzo, "Sub-band adaptive pre-equalized OFDM transmission," IEEE VTC, 1999. G.D. Durgin, Space-Time Wireless Channels, Prentice Hall PTR, 2002 Klaus Jostschulte studied Electrical Engineering at the University of Dortmund. After receiving his diploma degree in 1995 he joined the University's Circuits and Systems Lab, where he worked on quality improvement for images and image sequences. After receiving his Ph.D. in 2001 he became chief engineer at the Communication Technology Institute. His current research interests are networks for electronic media. Rüdiger Kays (M’ 03) received his diploma degree in 1981 and his Ph.D. in 1986 at the University of Dortmund. He was then with Grundig, Germany, where he was responsible for the company’s research and advanced development centre. In ’99 he became full professor at the University of Dortmund. His research topics cover signal processing of video information, multimedia and enhanced TV systems and networks for electronic media. Wolfgang Endemann studied Electrical Engineering at the University of Dortmund. After receiving his diploma degree in 1990 he joined the Communication Technology Institute at the University of Dortmund, where he worked on motion vector based video processing and image coding. He received his Ph.D. in 1998. His current research interests cover technologies and components of wired and wireless networks for audio/video applications.