This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Introduction• This work considered multicast video transmissions over WLANs.• WLAN unicast services do not scale to hundreds of fans, visitors or
customers in a contained area.• Multicast services are required, but these are fundamentally unreliable.• 802.11 provides reliable communications over unicast links via Automatic
Repeat Request (ARQ).• No standardized solution for reliable multicast. • Multicast packets are delivered as a simple broadcast service without
support for ARQ. Therefore, multicast transmission results in high packet loss.
Introduction• Video applications cannot tolerate higher packet loss rate.• Application Layer Forward Error Correction (AL-FEC) codes are possible
mechanisms of providing reliable multicast delivery without the need for the return channels.
• This paper considered a cross-layer design based on the latest Raptor Q (RQ) codes for transmitting high data rate video over the MIMO channels in realistic outdoor environments.
• In channels with high spatial correlation Spatial Multiplexing (SM) results in high packet loss.
• To address this issue, we explore a combination of SM multicast transmission with RQ codes.
• A 3D ray-tracer was used to model the channel matrix H between the AP and each user location.• A novel Received Bit mutual Information Rate (RBIR) abstraction technique is used to estimate the
Packet Error Rate (PER) statistics.• The video simulator can model the transmission of any H.264 video sequence over the MAC and PHY
layers of 802.11n.
3D ray tracing for a very large
dataset
Spatially & polarimetrically convolve ray data with measured
antenna patterns
+
Determine received packet trace
Determine PER for all MCS modes
Wideband channel frequency responseBit error rate (BER)
Packet error rate (PER)Bit level simulatorSISO downlink in AWGN
• The aim of the cross-layer design process is that for given mean channel SNR and H matrix determinant, select the optimum transmission scheme (SM or STBC), MCS mode m and Raptor code rate CR (if SM is selected) that provide the minimum total transmission time ,with the constraint that .
• A constant bit rate video sequence which consists of 1500 UDP packets was transmitted at 4 Mbps.
• RaptorQ source block length k=200 and symbol size T=1400 B.• Raptor code rate, CR, range: 0.• The transmission modes for an 802.11n 20 MHz channel profile (with a
400 ns GI) were used.• The peak rates: 144.4 Mbps for SM and 72.2 Mbps for STBC (with the
use of 64-QAM 5/6).• The application QoS requirement: <1%.• One Raptor symbol was placed into one UDP/IP packet.
no RaptorCR=0.95CR=0.9CR=0.85CR=0.8CR=0.75CR=0.7CR=0.65CR=0.6CR=0.55CR=0.5
Results: UDP PER performance with Raptor
10
• UDP PER performance of SM with respect to Raptor CR under different MIMO channel conditions: mean H matrix determinant 0.5 (left high correlation) and mean H matrix determinant 1 (right low correlation).
• Depending on the MCS, H matrix determinant and CR there would be as much as 8 dB improvement in the required SNR.
10 15 20 25 30 3510
-3
10-2
10-1
100
SNR (dB)
PE
R
QPSK 1/216-QAM 1/264-QAM 2/3
5 10 15 20 25 30 3510
-3
10-2
10-1
100
SNR (dB)
PE
R
10 15 20 25 30 3510
-3
10-2
10-1
100
SNR (dB)
PE
R
QPSK 1/216-QAM 1/264-QAM 2/3
5 10 15 20 2510
-3
10-2
10-1
100
SNR (dB)
PE
R
no RaptorCR=0.95CR=0.9CR=0.85CR=0.8CR=0.75CR=0.7CR=0.65CR=0.6CR=0.55CR=0.5
• Total transmission time versus SNR for H matrix determinant of 0.5 and corresponding optimum CRs.• For given mean SNR and H matrix determinant, the MCS mode and optimum CR pair that provide the
lowest total transmission time with the constraint that was determined.
• Transmission efficiency in terms of total transmission time for STBC and SM with different Raptor codes.
• CR=1 represents the results without Raptor code.
• STBC provides higher transmission efficiency than SM with CR = 0.5 at each MCS mode. Therefore, in the case of MIMO switching the code rate range must be CR > 0.5.
• For higher MCS modes there is little performance difference (gain) that can be exploited by Raptor codes therefore the available CR range is decreased.
• Optimum MIMO/MCS mode that provides most efficient data transmission for given mean SNR and H matrix determinant.
• When the correlation is high, STBC is selected until the SNR values that enable higher MCS modes to be implemented in SM, e.g. the required SNRs are 36 dB and 28 dB for the mean H matrix determinant of 0.1 and 0.5 respectively.
• When correlation is low, e.g. mean H matrix determinant 1, the use of Raptor codes allows lower MCS modes such as QPSK 3/4 and 16-QAM 1/2 in SM scheme to be selected.
• In areas with low SNR and high spatial correlation STBC shows better performance than SM. SM fails to provide service without Raptor codes (PHY rate is zero).
• SM takes advantages of Raptor codes, especially when correlation is low, and provides higher transmission efficiency, i.e. the overall channel occupancy time is less than the system without Raptor codes.
Conclusion• This paper presented a cross-layer design approach which considers Raptor
codes in the SM system as a means to enhance the reliability and transmission efficiency for transmitting high data rate video.
• A cross-layer simulator was used to investigate the optimum system parameters under different channel conditions and evaluate the performance in realistic outdoor environments.
• The algorithm implements mean SNR and H matrix determinant to adapt the system to changing channel conditions.
• We have shown that with the use of Raptor codes SM provides robust video transmission in harsh channel conditions and makes use of higher MCS modes when channel conditions are good therefore increases the overall transmission efficiency.