1. Evaluating Video Streaming over UWB Wireless Networks
Rukhsana Ruby
University of Victoria
Joint work with Yangyang Liu, and Jianping Pan

2. Outline Existing WLAN Technologies UWB Overview Evaluation Methodology
Performance Analysis
Performance Results
Discussion
Future Research

3. Existing WLAN Technologies
Support low data rate (11 to 54 Mbps)‏
802.11b and a/g (Achieve less than 50% of actual data rate)‏
Work in 2.4 and 5 Ghz frequency band.
avg throughput of g 10 Mbps
Contention – based MAC
Existing wireless technologies support very low data rate. For example b can support raw data rate up to 11Mps and a/g up to 54 Mbps. Apparently we can say these data rate is sufficient for high quality video streaming. However physical and mac layer overhead through these technologies we can only achieve 50% of data rate. IEEE devices work in the same 2.4 and 5 GHz unlicensed frequency bands as other home devices such as cordless phones and microwave ovens which causes interference and reduces SNR further. It is found that average throughput of g devices is around 10 Mbps. Contention based MAC also reduces achievable throughput due to channel idle and collision times.

4. UWB Overview UWB is a radio technology
UWB is regarded as the best technology for the high-speed wireless PAN. Why?
High speed at short range
Up to 480Mb/s currently. Ultimately support the speed at Gbit/s.
In the range of 10 meters.
Radio spectrum: 3.1 to 10.6 GHz (very large).
Low energy consumption level
The reasons why UWB is regarded as the best technology is due to its high data rate. It works 3.1 to 10.6 GHz frequency range to accommodate more device groups and thus supports such huge raw data rate. Since UWB devices consume very low power, uwb has very little interference with other devices.

5. UWB PHY DS-UWB MB-OFDM 28, 55, 110, 220, 500, 660, 1000, 1320 Mbps
MBOA: Multi-band OFDM Alliance; WiMedia
There are two camps of UWB, DS-UWB and MBOA-UWB. DS-UWB, referred to as Direct Sequence UWB, is based on Direct Sequence Spread Spectrum (DSSS) technology. MBOA-UWB, which eventually became WiMedia UWB, is based on the combination of Time-Frequency Coding (TFC) and Orthogonal Frequency-Division Multiplexing (OFDM) technology.

6. UWB MAC Time is divided into super frames.
256 MAS (Each MAS is 256us)‏
Beacon Period (First 32 MAS)‏
Contract and Expand able
Data Period
DRP and PCA (Like e)‏
Acknowledgement Policy
No, Block and Immediate Acknowledgement
Time is divided into super frames, each super frame is ms and is divided into 256 MAS where each MAS is 256 us. Super frame is again divided into beacon period and Data transmission period. Beacon period is usually first 32 MAS, however it’s contract and expand able.Two kinds of data transmission is possible during data transmission period of super frame. DRP period allows user exclusive hold of the channel, and it has also some variations; hard drp and soft drp. Another kind of data transmission of UWB PCA is just like EDCA. Three kinds of acknowledgement are supported in UWB MAC; No acknowledgement (receiver does not acknowledge sender if it receives data successfully), block (receiver send acknowledgement after receiving certain number of data), immediate acknowledgement (receiver sends ack just after receiving the data pkt)

7. Evaluation Methodoloy
Testbed Configuration
Video File Statistics

8. Testbed Configuration
uw1 and uw2
ui1 and ui2 for control access
Tzero ZeroWire Mini PCI 700 Revision B card with Dual antenna
Tea cans on antenna
RSSI -73 dBm
No PCA, only DRP
Collision if same MAS is reserved by uw1 and uw2
Two uwb devices with Tzero ZeroWire Mini PCI 700 revision B card with dual antenna in our lab approximately in 10 metres distance, we call them uw1 and uw2. Ethernet interface of these devices are called ui1 and ui2. In order to emulate a household environment full of obstacles and interference, we used two tea cans to cover uw2’s antenna for a resultant Received Signal
Strength Indicator (RSSI) around -73 dBm. In these two devices MAC layer support only DRP protocol, no PCA. However DRP is also restricted with limited functionalities-for example both uw1 and uw2 can reserve same MAS, collision happens when they try to transmit pkt in the same reserved MAS

9. Video File Statistics 4 MPEG-4 AVC video streams
Resolution 1920*1080, refresh rate 24 frames/sec
Average frame size kilo bytes
Average data rate is around Mbps
In our experiments, we used a two-minute high-definition video camera demo video clip as an example. The video has a resolution of 1920*1080 and refresh rate of 24 frames per second. We applied the MPEG-4 AVC reference encoder on the raw video. From the figure, we can tell the average frame size is about kilobytes, or the average data rate around Mbps. We used multiple video streams to fully utilize the UWB link, e.g., four streams to represent quad-HDTV scenarios for cinema-like experience.
Fig. Frame size vs. Frame Sequence Number

10. Performance Analysis Duration of the entire PLCP packet:
Maximum number of PLCP packets can go through in the reservation of n MAS is:
UDP pkt length
18 or 30 symbols us
12 symbols
12 us
Video streams are often transported in Realtime Transport Protocol (RTP), which is consequently encapsulated in UDP, IP and LLC protocol with 8, 20 and 16-byte header, respectively, or for an overall overhead of 44 bytes above the MAC layer. In WiMedia UWB, an OFDM symbol lasts TSYM = μs and can carry a different number of information bits depending on modulation and coding schemes, which jointly determine the physical layer data rate. In the Physical Layer Convergence Protocol (PLCP), a standard or burst PLCP preamble of Nsync = 30 or 18 symbols is pre-fixed for packet synchronization and channel estimation1, followed by a PLCP header of Nhdr = 12 symbols. The remaining of the PLCP packet including a 4-byte Frame Check Sequence (FCS), a 6-bit tail and extra pad bit lasts an integer multiple of six symbols, which depends on the number of information bits per six OFDM symbols (NIBP6S) at the given data rate. If the UDP payload length is L bytes, the duration for the entire PLCP packet over the air in μs is
10 us

11. Performance Results TxRate and Retry Limit (CBR and VBR)‏
Reservation percentage and pattern (CBR and VBR)‏

12. TxRate and Retry Limit (CBR)‏
This figure shows achieved goodput with different txrate and retry limit when the reservation pattern is 50% with reservation index 3. When the TxRate is increased, packet transmission time is reduced, but the PLR is increased. Therefore, when TxRate is slightly increased we see a considerable increase of achievable goodput. When TxRate is above 200 Mbps, the PLR is 100% and the achieved goodput is zero. This figure also shows that the increased retry limit actually reduces achievable goodput. Since the sender has to wait for the acknowledgement required for the acknowledgement, which keeps the channel idle for a while and eventually reduces goodput. The left table verifies the experimental result obtained with the analysed result.
Fig. Goodput vs. TxRate and Retry Limit

13. TxRate and Retry Limit (Video)‏
PSNR is the most well-known metric for video quality measurement. We sent video stream from uw1 to uw2 . This figure shows that the average PSNR is decreases with the increased TxRate due to FLR which is caused by PLR. The higher the PSNR value, the better the video quality. Link layer retransmission greatly improves PSNR. In fact, with a retry limit of 7, the PSNR can achieve almost 50 dB when TxRate is below 200 Mbps.
Fig. Average PSNR vs. TxRate and Retry Limit

14. Reservation percentage and pattern
Non-allocatable MAS
Allocatable MAS
This is the picture of entire superframe in our uwb boxes. There are 16 blocks in the above picture. Each block is the representative of 16 MASs. In our uwb boxes, we cannot allocate first 32 and last 16 MASs which is shown in the figure.
Entire Super Frame

15. Reservation percentage and pattern
32-alternative 5
64-alternative 6
16-alternative 4
8-alternative 3
4-alternative 2
2-alternative 1
Reservation Pattern
Reservation Index
To see the effect of different reservation pattern, we reserve 50% of total allocatable MASs in various patterns. In the table, I have shown diff reservation patterns and have indexed them in numbers from 1 to 6. This figure shows the sample 16 MAS reserved when reservation pattern index is 1.
Reserved MAS
Un-reserved MAS
Sample of 16 MAS for reservation pattern index 1 & 3

16. Reservation percentage and pattern (CBR)‏
This figure shows the achievable throughput with different retry limit and reservation pattern at 53.3 Mbps. When the reservation becomes clustered, due to reduced turnaround overhead and guard time, the achievable throughput is increased. Again due to the time for acknowledgement, the achievable throughput is reduced with a higher retry limit. The table in the right side compares the throughput obtained by experimentation with the calculated one, we see both numbers are almost same.
Fig. Throughput vs. Reservation Pattern

17. Reservation percentage and pattern (CBR)‏
These two figures show the packet delay affected by service interval when reservation index are 1 and 5 respectively. Left figure is when the every alternate 2 MASs are reserved. Transmission and Receiving time curves are shifted horizontally to align with the start of a super frame at packet sequence number 104. The vertical difference between Rx and Tx curve indicates roundtrip time which is considered as one way delay echo pkt comes from receiver to sender over ethernet interface. Due to the offered load is higher than this pattern can sustain, we see an increased packet delay due to the increased queuing delay. In the right hand side figure, when one alternate 32 MAS are reserved, there are more packets served in the same interval, indicating higher throughput; however it suffers more obvious service interval than the other papern.
Fig. Packet Delay vs. Packet Sequence Number

18. Reservation percentage and pattern (Video)‏
To further identify the effect of different reservation pattern on service rate and interval, we show the max accumulated jitter for sample video file. Frame jitter is defined as the time difference to send two consecutive video frames. When the reservation becomes clustered, due to the increased service interval, the MAJ is increased.
Fig. Frame jitter vs. Reservation Pattern

19. Discussion Tradeoff between TxRate and Retry Limit
Throughput, Latency tradeoff between clustered and scattered reservation.
There are two tradeoffs we have got from our experimentation. A higher TxRate implies higher packet loss ratio at the same received SNR, and a higher retry limit can reduce packet loss but may reduce goodput. with a clustered reservation, there are fewer turnarounds in a superframe, which leads to less guard time and higher channel utilization. On the other hand, for the same number of reserved slots, a clustered reservation will increase service
interval, which potentially increases access latency. Therefore, in order to increase throughput and reduce latency,we have to strike a balance between scattered and clustered reservations, particularly for video traffic.

20. Future Research Model UWB MAC
Centralized algorithm for MAS reservation taking account of service interval.
Distributed algorithm in beacon period for MAS reservation.

21. Thank you!