Project: IEEE P802.15 Working Group for Wireless Personal Area Networks (WPANs)

1. Project: IEEE P802.15 Working Group for Wireless Personal Area Networks (WPANs) Submission Title: [MAC Performance enhancements for Alt-PHY] Date Submitted: November 13, 2002 Source: [Chuck Brabenac] Company: [Intel] Address: [2111 NE 25th Ave, Hillsboro, OR 97124] Voice: [503-264-8575], FAX: [503-264-3483], E-Mail: [chuck.brabenac@intel.com] Re: [02/105r20 clause 4] Abstract:[Illustrates and justifies the need for various MAC performance enhancement modifications to support the 802.15.3a alt-PHY] Purpose: [To set context for discussion of 02/105r20 clause 4] 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 these viewgraphs becomes the property of IEEE and may be made publicly available by P802.15. Chuck Brabenac, Intel R&D

2. Overview: 15.3a MAC performance enhancement need • 15.3 MAC currently supports <=55Mb/s PHY • Its throughput efficiency can be very high • 15.3a PHY will be 110, 200 & scalable to 480Mb/s+ • On the order of 10x+ PHY rate increase • Reasonable to expect MAC performance enhancements to preserve throughput efficiency • 15.3a CFA responses cited heavy MM needs • MPEG-2, MPEG-4 movies, etc. • We should assure the MAC can carry this traffic efficiently & reliably Chuck Brabenac, Intel R&D

3. Simple “best case” CFP analysis at higher rates (1) • PHY parameter assumptions (pure fiction for now) • 20us preamble (02/105r20 target) • 10us SIFS, 2us MIFS (from 15.3 D14-pre) • 22 Mb/s base rate (PHY & MAC headers, etc.) • 2K byte packets • Error-free channel • Application/flow assumptions • Single continuous flow, consuming nearly entire SF • Ignored beaconing, guard time, tail symbols, stuff bits • Calculations made at 55,110,200,480 and 1000 Mb/s: • Immediate ACK policy • Delayed ACK policy on 16 packets w/ SIFS • Delayed ACK policy on 16 packets w/ MIFS • Delayed ACK policy on 16 packets w/ MIFS and no preamble on subsequent frames of a burst Chuck Brabenac, Intel R&D

4. Simple “best case” CFP analysis at higher rates (2) Chuck Brabenac, Intel R&D

5. Simple “best case” CFP analysis at higher rates (3)Some Observations • Delayed ACK/MIFS developed by TG3 team helps significantly (~20% @480Mb/s) • PHY preamble overhead remains a dominant factor to deal with at higher rates (~22%) • To scale upward, we should consider reducing/eliminating it on delayed ACK bursts Chuck Brabenac, Intel R&D

6. Improving multimedia performance • Minor changes can reduce job failure rates of multimedia flows (e.g., MPEG-4) • Flow queue depth field could be added, enabling more effective scheduler (SRPT-F, etc.) • Effectively increases capacity of the MAC for this type of dynamic flow rate traffic (see next foil) • Refer to 02/297r1 QoS performance analysis and simulation results for more info Chuck Brabenac, Intel R&D

7. 70% Average performance improvement over 802.15.3 MAC ~95% Channel utilization Improving multimedia performance Chuck Brabenac, Intel R&D

8. Summary • 15.3a MAC performance enhancements are necessary as PHY data rates scale upward • Justifies needed MAC modifications as per 02/105r20, clause 4 Chuck Brabenac, Intel R&D