doc.: IEEE /0910r0 SubmissionLaurent Cariou, OrangeSlide 1 Carrier-oriented WIFI for cellular offload Date: Authors: July 2012

doc.: IEEE /0910r0 SubmissionSlide 2 Abstract Presentation for WNG SC July 2012 July 2012 Laurent Cariou, Orange

doc.: IEEE /0910r0 SubmissionSlide 3 Wi-Fi is increasingly becoming an integral component in the delivery of broadband wireless services by fixed and mobile network operators around the globe Hotspot deployments will increase strongly in the next few years –to support the capacity crunch, cellular operators will need to densify their networks and the small cells deployments will be inevitable, for indoor but also outdoor. –3GPP integrated solutions exist (micro, pico, femtocells) but WIFI has a role to play in such deployments assuming the fact that it fulfills the requirements –3GPP small cells should arrive in the market in 2014, WIFI should be competitive at this time Background July 2012 Laurent Cariou, Orange

doc.: IEEE /0910r0 SubmissionSlide 4 We believe that hotspot requirements are different from home and entreprise requirements –higher number of users with lower throughputs –strong performance requirements in instable environments with interferers –different market performance evaluation: average throughput and cell-edge throughput more important than peak throughput Following n, ac have focused its energy on very high throughput scenarios and provided technical solutions accordingly –home network and entreprise markets will largely benefit from this WIFI generation –in the meanwhile, the hotspot market, especially for cellular offload, will not benefit as much from the evolution of ac, while it is a major growing market Background July 2012 Laurent Cariou, Orange

doc.: IEEE /0910r0 Submission Different hotspot deployment scenarios Slide 5 Classical hotspots using a fixed network interconnected to the cellular core network (high layer interconnection) sometimes (not always) deployed with a controller Hotspots deployed within LTE small cells WIFI directly connected to the cellular core network (low layer interconnection) Home private APs used as hotspots Private SSID for home users and public SSID for cellular offload July 2012 Laurent Cariou, Orange

doc.: IEEE /0910r0 Submission Different hotspot deployment scenarios Slide 6 Classical hotspots using a fixed network interconnected to the cellular core network (high layer interconnection) stand-alone APs cluster APs with a controller WIFI controller Light AP 3GPP gateway Proprietary coordination through WIFI controller No coordination with other APs July 2012 Laurent Cariou, Orange

doc.: IEEE /0910r0 Submission Different hotspot deployment scenarios Slide 7 Hotspots deployed within LTE small cells (candidate for LTE-A rel12) Better integration of WIFI in the RAN Eventually, WIFI directly connected to the cellular core network (low layer interconnection controlled by LTE) July 2012 Laurent Cariou, Orange Macro-cell 3G/4G Small cell WIFI

doc.: IEEE /0910r0 SubmissionSlide 8 Home private APs used as hotspots Private SSID for home users and public SSID for cellular offload Private Public Different hotspot deployment scenarios WIFI gateway/controler 3GPP gateway Home APs used as Hotspots Coffee shop with many public APs available at the cell edge July 2012 Laurent Cariou, Orange

doc.: IEEE /0910r0 SubmissionSlide 9 Improve average throughputs –requires the improvement of cell-edge (BSS-edge) throughput –requires the reduction of the degrading impact of cell-edge STA on other STA Improve the scheduling capabilities of the AP to apply fairness strategies –note that this will also improve the average throughput –our objective: building a consensus to ensure that such solutions will reach the market Improve robustness to interference (mostly OBSS management) –better coordination is needed between co-located BSSs for spectrum sharing, interference management and QoS –better robustness to band saturation Ensure the adaptation to outdoor deployments –potential range extension, support for longer delay spreads Major requirements July 2012 Laurent Cariou, Orange

doc.: IEEE /0910r0 Submission Improve average throughput by improving cell-edge stand-alone APs Operators will deploy many stand-alone APs as small cells. the objective of small cells is to offload as much traffic as possible from the overlaying macrocell –the strategy is therefore that all STAs in the coverage of the small cells connect to it (small cell range extension in LTE-A) –if cell-edge users have a too strong impact on the performance of the whole WIFI cell, this strategy is not applicable. We therefore need to find solutions for this. Cell range-extension Macro-cell Small cell July 2012 Slide 10Laurent Cariou, Orange

doc.: IEEE /0910r0 Submission Improve average throughput by improving cell-edge APs cluster Operator will also deploy clusters of APs with efficient coverage/capacity. current WIFI strategy is to densify the number of APs, with strong coverage overlap. In these cases, the use of low MCSs can even be forbidden –This allows to avoid bad performance at cell-edge and the throughput degradation impact of cell-edge user on all other users Compared to enterprise deployments, outdoor hotspot deployments will have strong positioning constraints On top of that, many WIFI hotspots will be deployed as a small cell which incorporates a WIFI AP and an LTE pico/femtocell –The deployment strategy of such small cells will mostly be based on LTE pico/femto which will usually have a slightly higher coverage which don’t push for densifying as it creates interference  Finding technical solutions to improve cell-edge performance can not be avoided in hotspot deployments as it is in enterprise deployments  Efficient SON mechanisms are required for capacity/coverage optimisation, handover optimization July 2012 Slide 11Laurent Cariou, Orange 3G/4G Small cell WIFI

doc.: IEEE /0910r0 Submission In many cases, especially in the home APs hotspots case, STAs will be far from APs but will see many APs. –The throughput of ths cell-edge STA will be limited to reduce the impact on private SSID –such a STA should be able to associate on multiple APs/hotspots in order to aggregate the throughputs offered by all APs –this could also facilitate handovers Slide 12 WIFI gateway/controler 3GPP gateway Home APs used as Hotspots Coffee shop with many public APs available at the cell edge Improve average throughput by improving cell-edge Home APs July 2012 Laurent Cariou, Orange

doc.: IEEE /0910r0 Submission Potential technical solutions: –Beam selection mechanisms at AP side, especially for the weaker UL signal –Longer range lower throughput transmission scheme, associated with a multiple access scheme to keep a low occupation of the channel and a good MAC efficiency: OFDMA is a good candidate –Coordination between neighboring APs adaptation of CoMP mechanisms defined in LTE-A: multi-cell high-level scheduling Multiple MAC (MMAC) association –… Improve average throughput by improving cell-edge July 2012 Slide 13Laurent Cariou, Orange

doc.: IEEE /0910r0 SubmissionSlide 14 Currently, EDCA leads to throughput fairness –leads to bad average throughput, high airtime occupation by cell-edge users, … We are looking for ONE realistic solution to enable different fairness strategies. Bringing scheduling capabilities to the AP is a solution for this. Some technical solutions have already been proposed, none of them have reached the market. We are not pushing for a particular technical solution but want to reach a consensus to ensure it will reach the market. It can be –based on EDCA with parameterization improvement –based on EDCA with the use of a multiple access scheme like OFDMA or simple multi- user aggregation, following the example of ac MU-MIMO (scheduling capabilities for the AP) –based on EDCA with protocol modifications adapting strategies proposed in ah –… or based on modifications of HCCA Improve the scheduling capabilities of the AP to apply fairness strategies July 2012 Laurent Cariou, Orange

doc.: IEEE /0910r0 Submission Improve robustness to interference Self organizing networks (SON) is a very important feature of LTE, in order to save OPEX and improve capacity/coverage –It enables self-configuration of new cells, automatic neighbor relation (ANR), load balancing, mobility robustness/handover optimization, Energy savings, Capacity and coverage optimisation (Radio ressource management (RRM) to detect dead spots, reduce interference between cells, select channels/bandwidth, adjust tilt and MIMO antenna configurations, adjust transmission power, …) For WIFI, SON mechanisms (especially RRM) can easily be done with proprietary solutions on an area if all APs are managed by a controller (entreprise scenario). Products already exists. But WIFI Hotspots deployments will be done by using -multiple stand-alone APs potentially from different vendors (example of home APs hotspots) -hotspots clusters deployed by the operator, by other operators or by a tier who deployed first with a roaming license In such cases (multi-controller, multi-operator, multi-vendor), SON mechanisms require standardized features like exchange of information between neighboring APs –802.11ah can be a good candidate for AP-to-AP coordination exchange: we need to define these information exchange July 2012 Slide 15Laurent Cariou, Orange

doc.: IEEE /0910r0 Submission Ensure the adaptation to outdoor deployments WIFI has been designed for indoor environments. –OFDM parameters has been set to cope with small delay spreads When used in a outdoor environment, the performance can be severly degraded –OFDM is not well parametrized (some papers show that performance of 11g is worth than 11b) delay spread excess cyclic prefix which cause interference, errors due to too short preambles lead to packet loss. Outdoor WIFI hotspots (especially within 3G-4G small cells) will be heavily deployed in the incoming years if the performance is preserved –we can modify OFDM parameters or apply downsampling techniques as in 11af or 11ah. July 2012 Slide 16Laurent Cariou, Orange

doc.: IEEE /0910r0 SubmissionSlide 17 IEEE should provide solutions to address this growing market. What sort of thing are we looking for? –A new study group, perhaps “Carrier-oriented WIFI - Cellular offload”, with the objective to: improve cellular offload WIFI performance for hotspot deployments –We envision PHY and MAC light modifications in order to improve current WIFI generation for hotspots requirements (not a new generation) n/ac as the basis of such improvement (2.4 and 5GHz) ah as a good candidate for AP-to-AP coordination exchange –Different levels of improvements: depending on the targeted deployment scenario: home APs, hotpots deployed by operators and WIFI hotspots aggregated with LTE and on the WIFI technologies integrated in chipsets: - 11n/ac in STAs and APs - 11n/ac in STAs and 11n/11ac/11ah in APs - 11n/ac/ah in STAs and 11n/11ac/11ah in APs Proposal July 2012 Laurent Cariou, Orange

doc.: IEEE /0910r0 SubmissionSlide 18 Should IEEE consider the creation of a study group to further discuss the topic of “cellular offload WIFI” ? –Yes –No –Need more information –Abstain/Don’t care Straw Poll July 2012 Laurent Cariou, Orange

doc.: IEEE /0910r0 SubmissionSlide 19 Request approval by IEEE 802 LMSC to form an Study Group to consider the cellular offload WIFI / Carrier-oriented WIFI [as described in doc ] with the intent of creating a PAR and five criteria. Moved:, Seconded:, Result: y-n-a Motion to create a Study Group July 2012 Laurent Cariou, Orange

doc.: IEEE /0910r0 SubmissionSlide 20 Annexes July 2012 Laurent Cariou, Orange

doc.: IEEE /0910r0 SubmissionSlide 21 Evaluation methodology for an improved fairness –inspired from 3GPP evaluation methodology: peak throughput, average throughput and cell-edge thoughput (5%) The cell edge user throughput is defined as the 5% point of CDF of the user throughput normalized with the overall cell bandwidth. Average spectrum efficiency is defined as the aggregate throughput of all users (the number of correctly received bits over a certain period of time) normalized by the overall cell bandwidth divided by the number of cells. The average spectrum efficiency is measured in b/s/Hz/cell. Small cell performance evaluation July 2012 Laurent Cariou, Orange

doc.: IEEE /0910r0 Submission Efficient share of spectral resources between BSSs Each AP signals its occupation and load at 5GHz enables dynamic channel selection without scanning benefit from 11ah range for a better knowledge of channel occupation Potential negociation between APs using signaling band for channel occupancy dynamic channel occupancy to satisfy the needs from all APs AP need: 80MHz 60% load  co-channel with AP2 AP need: 80MHz 90% load  co-channel with AP2 not possible  channel switch request to AP2 July 2012 Slide 22Laurent Cariou, Orange

doc.: IEEE /0910r0 Submission Efficient share of spectral resources within a BSS STAs far from AP (cell-edge STA) can strongly impact the performance of other APs transmit with low MCS and low bandwidth for a long time One solution is to enable to simultaneously transmit on the primary 20MHz channel towards this cell-edge STA and on secondary channel towards other STA (simple multi-band DL-OFDMA) toward STA1 toward STA2 Primary channel Secondary channel July 2012 Slide 23Laurent Cariou, Orange

doc.: IEEE /0910r0 Submission Use of ah as a signaling band for improved cooperation Use ah for signaling and ac/ad/af for data transmissions Information sharing between APs to improve cooperation/coordination between neighboring cells –interference management, frequency planning, power control, … Information sharing between AP and STAs to improve quality of service, energy efficiency, interference management Slide 24 July 2012 Laurent Cariou, Orange