November, 2007 Project: IEEE P802.15 Working Group for Wireless Personal Area Networks (WPANs) Submission Title: [Superframe Sizing Procedure for Uncompressed Video Traffic] Date Submitted: [November 10, 2007] Source: [Wooyong Lee1, Jinkyeong Kim1, Yongsun Kim1, Kyeongpyo Kim1, Hyoungjin Kwon1, Kyungsup Kwak2, Seokho Kim2, Xizhi An2, Saurabh N. Mehta2, Zhiquan Bai2, Sangkyoon Nam2] Company: [ETRI1, Inha University2] Address: [ETRI, 161 Gajeong-dong, Yuseong-gu, Daejeon, 305-700, Republic of Korea]1, [6-141B, Inha University, 253 Yonghyun-dong, Nam-gu Incheon 402-751, Republic of Korea]2 Voice: [], FAX: [], E-Mail: [wylee@etri.re.kr1, jkkim@etri.re.kr1, doori@etri.re.kr1, kpkim@etri.re.kr1, kwonjin@etri.re.kr1, kskwak@inha.ac.kr2, sylvstar@inhaian.net2, anxizhi@inhaian.net2, sdzqbai@inhaian.net2, smehta@inhaian.net2, sknam@inhaian.net2] Re: [] Abstract: [To propose a sizing method to find a proper superframe size] Purpose: [To be considered in IEEE 802.15.3c standard] Notice: This document has been prepared to assist the IEEE P802.15. It is offered as a basis for discussion and is not binding on the contributing individual(s) or organization(s). The material in this document is subject to change in form and content after further study. The contributor(s) reserve(s) the right to add, amend or withdraw material contained herein. Release: The contributor acknowledges and accepts that this contribution becomes the property of IEEE and may be made publicly available by P802.15. ETRI

Overview Memory buffer size in receiver is very important. November, 2007 Overview Memory buffer size in receiver is very important. Superframe size is related to the buffer size. The application of uncompressed video transmission is delay sensitive and requires constant bandwidth allocation. Allocating Bandwidth below the required throughput causes service failure. So, PHY-SAP payload bit rate is key parameters Simulation and analytical results show that we can find superframe size and payload length according to the requirements (data rate, delay and memory usage). Propose a sizing procedure to find a proper superframe size according to the requirements of uncompressed video transmission. ETRI

802.15.3c Performance Metrics Throughput End-to-End Delay November, 2007 802.15.3c Performance Metrics Throughput measured in terms of bits per second, is the amount of data delivered successfully by the peer MAC-SAP. End-to-End Delay measured in terms of second, is the amount of time taken for a MAC SDU to be transferred from the MAC-SAP of the transmitter to the peer MAC-SAP of the receiver. Memory usage (MAC) Memory usage represent the minimum buffer size required for continuous video representation ETRI

Uncompressed Video Streaming November, 2007 Simulation Model (UM1) PNC Control/commands Beacon Uncompressed Video Streaming DEV-0 DEV 1 LRT (1.485 Gbps) Common mode (48.5 Mbps) Uncompressed Traffic Model :1080p 30f 20b - 2200(1920+280)*1125(1080+45)*30*20 = 1.485 Gbps - CBR traffic -> 44000bit (2200*20) / 29.63us ETRI

CoMPA’s Superframe Structure November, 2007 CoMPA’s Superframe Structure BP: Beacon Period CAP: Contention Access Period CTAP: Channel Time Allocation Period GT: Guard Time PLCP: Physical Layer Convergence Protocol ETRI

Considering Superframe Structure November, 2007 Considering Superframe Structure ETRI

Beacon Frame with CTA IE November, 2007 Beacon Frame with CTA IE * Note 1: It shows the minimal length of beacon frame with CTA IE included. Note 2: In this example, there is one Channel Time Allocation (CTA) IE that contains one CTA block assigning the channel time for one traffic flow. * PLCP header (17 octets) = PHY header (5 octets) + MAC header (10 octets) + HCS (2 octets) ETRI

Preamble and PLCP Header November, 2007 Parameters Assumed Use common mode beacon to 47.8 Mbps Assume error free PHY channel Set video traffic to CBR traffic mode One MAC SDU is composed of data bits transmitted in one horizontal line. Assume no buffer delay due to higher data rate Assume no propagation delay due to short distance ( < 10 meter) Consider transmission delay in terms of Overhead & Payload Size Simulate in the NS-2 Parameters Value Superframe Size 1 msec ~ 64 msec Beacon Interval 15..344 usec CAP Duration 0 / 200 usec Guard Time 0.02 usec SIFS 2.5 usec Preamble and PLCP Header 8.157 usec PHY-SAP Rate 1530 Mbps Payload Size 2 KB ~ 64 KB Sub-header Size 4 Bytes ACK Policy N/A Aggregation Method Applied MAC-SDU Size 5.5 KB ETRI

November, 2007 Analytical Results (1) ETRI

November, 2007 Analytical Results (2) ETRI

November, 2007 Simulation Results (1) Performance of video transmission, without CAP, PHY-SAP rate = 2000 Mbps ETRI

November, 2007 Simulation Results (2) Performance of video transmission, with CAP, PHY-SAP rate = 2000 Mbps ETRI

November, 2007 Simulation Results (3) Performance of video transmission, without CAP, PHY-SAP rate = 1632 Mbps ETRI

November, 2007 Simulation Results (4) Performance of video transmission, with CAP, PHY-SAP rate = 1632 Mbps ETRI

Procedure of Sizing the Superframe Size November, 2007 Procedure of Sizing the Superframe Size ETRI

An Example for setting 1080p30f20b video (1.485 Gbps) November, 2007 An Example for setting 1080p30f20b video (1.485 Gbps) (with CAP = 200 us) PHY-SAP rate (Mbps) Delay Bound (us) Payload length (kB) Proper Superframe size (ms) 2000 < 600 60.5 2 < 400 60.5 4 < 200 22 8 1632 < 600 60.5 8 < 400 44 16 < 200 N/A N/A Note: 1) According to the data rate and delay requirements, a groups of candidate <superframe size, payload> settings can be selected; 2) Then, according to the calculation of CTAP efficiency, as large as possible payload size tends to be selected to obtain higher efficiency. ETRI

Consideration of Buffer and Its Utilization November, 2007 Consideration of Buffer and Its Utilization The buffer limit would be one of capability parameters of the device. It shows how many data can be temporarily stored in the device for the smoothly video representation. The occupation of this kind of buffer is related with data rate, and superframe size, etc. i.e. Minimal buffer required = data rate * superframe size Given a buffer limit, however, we would want to fully utilize it in order to achieve better performance. It is suggested that within the buffer limit of the device, select as large as possible superframe size in order to increase the CTAP efficiency additionally and reduce the delay. ETRI

November, 2007 Conclusions With this comparison we can propose the minimum requirement on PHY-SAP rate to transfer uncompressed video traffic. To improve the transmission efficiency, the aggregation should be used, which results in larger payload length. However, the transmission delay is correspondingly increased. By considering the expense of beacon period and CAP, the channel time for each flow and relevant superframe size should be lengthened. However, there would be more buffered data and then more memory would be required. The superframe size should be set to relative bigger to achieve proper effective throughput, when the PHY-SAP rate is low. On the other hand, for the high PHY-SAP rate, the superframe size could be shortened to lower the memory requirement. In both cases, the delay should be considered in the selection of superframe size and payload length if the QoS is emphasized. Our proposal provide the baseline for finding superframe size in terms of delay requirement and memory usage. ETRI

November, 2007 Appendix ETRI

Throughput Calculation November, 2007 Throughput Calculation Above formula is valid only for 1-1 Usage model, for other cases, formula will be changed a bit. ETRI

Delay and Throughput Definitions November, 2007 Delay and Throughput Definitions Maximum achievable throughput represents the maximum MAC capacity. In real world scenario, we can’t achieve the maximum achievable throughput because of several reasons like MSDU size, collision , etc. Minimum Delay bound is the minimum transmission time required by a MAC MSDU to reach destination MAC layer from Source MAC layer ETRI

Frame Structure Basic Aggregation Mechanism and frame structure November, 2007 Frame Structure Basic Aggregation Mechanism and frame structure ETRI

Parameters on Video Resolution November, 2007 Parameters on Video Resolution Formant V Freq HRES VRES DE_CNT DE_LIN Rate 24bit Rate 20bit Hsync (sec) Vsync (sec) 1080i 30 2200 562/ 563 1920 1080 5760 4800 3.77104E-06 0.000652/ 0.000681 1080p 1125 0.001333 60 1.88552E-06 0.000667 ETRI