Below 6GHz 11vht PAR scope and purpose discussion March 2008 doc.: IEEE 802.11-08/0219r4 March 2008 Below 6GHz 11vht PAR scope and purpose discussion Date: 2008-04-17 Authors: Courville (Motorola), Kraemer (Marvell), Engwer (Nortel), De Vegt (Qualcomm), Erceg (Broadcom) Marc de Courville (Motorola)
March 2008 doc.: IEEE 802.11-08/0219r4 March 2008 PAR: scope The scope of this project is to define an amendment that shall define standardized modifications to both the 802.11 physical layers (PHY) and the 802.11 Medium Access Control Layer (MAC) so that modes of operation can be enabled that are capable of supporting: A multi-user BSS peak aggregated throughput of at least 1Gbps as measured at the MAC data service access point (SAP) Shall the constraint be applied only to the AP (non always existing entity: IBSS)? Change to BSS or IBSS capacity (ambiguity between different types e.g. Shannon etc.)? Specifying aggregated or multi-user is necessary in the IBSS case whereas for the BSS case we could apply the metric to the AP Shall we revert to a per link metric? Proposal: A BSS or multiuser IBSS throughput of at least 1Gbps as measured at the MAC data service access point (SAP) Capability of handling multiple communications at the same time Below 6GHz carrier frequency operation excluding 2.4GHz operation and ensuring backward compatibility and coexistence with legacy IEEE802.11a/n devices in the 5GHz unlicensed band. Courville (Motorola), Kraemer (Marvell), Engwer (Nortel), De Vegt (Qualcomm), Erceg (Broadcom) Marc de Courville (Motorola)
PAR: purpose Change To: March 2008 doc.: IEEE 802.11-08/0219r4 March 2008 PAR: purpose Change The purpose of the project is to improve the 802.11 wireless local area network (LAN) user experience by providing significantly higher aggregated throughput for existing WLAN application areas and to enable new market segments for operation below 6 GHz including distribution of multiple multimedia/data streams. To: The purpose of the project is to improve the 802.11 wireless local area network (LAN) user experience by providing significantly higher BSS or multiuser IBSS throughput for existing WLAN application areas and to enable new market segments for operation below 6 GHz including distribution of multiple multimedia/data streams. Courville (Motorola), Kraemer (Marvell), Engwer (Nortel), De Vegt (Qualcomm), Erceg (Broadcom) Marc de Courville (Motorola)
Rate expectation discussion March 2008 Rate expectation discussion Preliminary remarks/questions: One could think that we must go far beyond 11n max PHY rate but is this realistic for a reasonable client configuration/cost? 11a max PHY rate is achievable at (54Mbps) whereas 11n max PHY rate (600Mbps) has not been demonstrated yet: this need to be considered as a specification number. Initially MAC SAP throughput was introduced for providing a better metric for setting users expectations but currently peak PHY throughput is still used as the marketing number on the product boxes: shall we revert the policy and come back to PHY throughput metric to define the PAR? Some baseline reference numbers: 11a max PHY rate: 54Mbps, estimated MAC SAP rate (without aggregation, 5 users): 27Mbps (2x) 11n: max PHY rate (40MHz 4STS Short GI): 600Mbps, PAR MAC SAP target rate: min max of 100Mbps Need to anticipate that this method could be used for PAR MAC SAP throughput expectation derivation 11a->11n PAR gap: 100/54=1.85 11n->11vht gap should thus be to match evolution rate: PAR(11vht)=PAR(11n)/max(11a)*max(11n)=1.85*600=1.11Gbps! Reality sanity check: some rough initial MAC SAP throughput numbers extrapolated from 11n (5 users contending, 2ms TxOP AMSDU aggregation) 3ss 4x20MHz: 680Mbps, 3ss 5x20MHz: 850Mbps 2ss 4x20MHz: 450Mbps, 2ss 5x20MHz: 560Mbps Question: limit cost of client to 2ss? Reference 11n PHY max rate: 2ss: 130/270Mbps, 3ss 195/405Mbps for 20/40MHz Proposed methodology for the peak aggregated BSS throughput derivation Increase moderately PAR MAC SAP target increase evolution of 3.7x from 11a->11n to 5x for 11n->11vht: i.e. 500Mbps Concluding remark: a reasonable PAR target will allow not to put too much constraints on entry point products and does not preclude defining multiple Gbps modes following the model introduced in 11n. The gap that is recommended to consider is on the MAC SAP rates evolution Courville (Motorola), Kraemer (Marvell), Engwer (Nortel), De Vegt (Qualcomm), Erceg (Broadcom)