white paper Street Smarts: Why Outdoor Wi-Fi success hinges on adaptive radio technology

Transcription

1 white paper Street Smarts: Why Outdoor Carrier Wi-Fi network deployments to-date reveal predictable trends and consistent priorities: great connectivity; standards-based integration into existing back-end environments; seamless user and operator experiences; and an agile deployment architecture that enables rapid, efficient service innovation. In this paper, we re tending to the first and usually most important requirement, particularly for outdoor environments. Great connectivity is task #1, without which nothing else matters. Outdoors, this hinges on reliable, high-speed client connections while Handling high user counts per radio Adapting to interference and changing conditions Solving site acquisition and small-cell backhaul challenges Efficiently integrating with licensed-band small cells, including accommodation of neutral-host requirements For more details, read about the Ruckus SmartCell Architecture The Nature of the Problem Getting back to the basics of the problem why outdoor Wi-Fi is needed in the first place high-density user demand for mobile and nomadic Internet service is outstripping the capacity of macro and micro cellular networks deployed today. According to forecasts, the data growth trajectory will create more disparity between capacity supply and demand, increasing service provider pressure in most markets in the next few years. But the capacity problem is not evenly distributed. The kind of demand density that can be economically served with small cells (licensed, unlicensed, or both) appears outdoors in dense urban environments, such as on busy streets.

2 A collection of challenges makes these environments very tough on network operators: High user density, naturally Heavy bias toward the least capable mobile clients (i.e. smartphones) Few places to put radios, limited backhaul options, and tight constraints on mounting assets Concentration of demand along corridors and in other spaces defined by the city streetscape Spectrum congestion and interference from both unlicensed Wi-Fi networks and non-wi-fi sources Constantly changing conditions (people, vehicles, weather, etc.) LoS issues make spectrum reuse more difficult NLoS issues can exacerbate backhaul challenges and related cost minimization Licensed small-cell integration with macro networks In the next sections, we ll examine how two different classes of carrier Wi-Fi fare in addressing these many challenges. Two Different Classes of Carrier Wi-Fi There are two general types of Wi-Fi. The first is based on adaptive technology (BeamFlex, ChannelFly, etc.) with innovations FIGURE 1: Density Demand: Heatmaps showing concentrations of user density demand. Source: Keima, keima.co.uk and investment in both RF and MAC optimizations, available from Ruckus called Smart Wi-Fi. The other is based on conventional reference design (CRD) implementations, offering a basic RF design with unpredictable connectivity and capacity in challenging environments. Substantial differences exist along six key feature dimensions, as described below. FEATURE SMART WI-FI CONVENTIONAL REFERENCE DESIGNS Access signal optimization Adaptive polarization diversity Adaptive channel selection Mesh backhaul approach Integration with licensed-band small-cell radios Packaging *For additional reading: Using all the Tools You Can Continuous, automatic smart antenna tuning from among thousands of possible antenna patterns, packet by packet and client by client, based on mature, fieldproven algorithms (does not require client software support)* Included benefits mobile devices and RF environments with unpredictable antenna orientations and polarizations Based on measured, realized capacity, using every available channel across the network Leverages both adaptive antenna and smart mesh technology to self-organize and select best path links, optimize throughput, and maintain robust performance across changing conditions Flexible, modular approach, integrated at deployment time with support for external nodes Highest performance to weight and volume ratios, with integrated adaptive antennas and a range of coveragepattern options Manual tuning at deployment time, then fixed configuration. DSP-based transmit beamforming is advertised, but requires client-side support, and real-world measurements show no benefit to client access. Not available mobile devices experience unstable performance as environment and orientation change Based on signal strength guesswork (not highly correlated with real capacity) and usually limited to nonoverlapping channels Limited self-organizing behavior, limited success in adaptive reconfiguration (if any); individual link optimization (TxBF) incompatible with other MIMO modes; antenna design is static for a single pre-determined link Embedded approach, integrated in manufacturing, limiting licensed equipment vendor choice Usually 2x higher weight and size, with external antennas in most cases page 2

3 Large Numbers of Low-Bandwidth Clients Our goal is to satisfy the data needs of heavy concentrations of users in an urban environment. Current demand density can already approach, and occasionally exceed, 100 users per access node. There are several inherent challenges with such high densities. Highly mobile outdoor users are most likely to use devices and create demand density in waiting areas (e.g. public transportation stations/platforms) or indoor dwell spots like restaurants, or while in transit on public transport; naturally, the device type breakdown will heavily favor smartphones. Such devices are typically low-powered with physically constrained antenna design, and support only single-stream n; even under the best conditions, bandwidth for these devices can mirror that of g. As a result, AP capacity can be limited by the capabilities of the supported client population. It becomes critically important that the maximum capability be wrung out of each client connection as well as the WLAN system as a whole. There are substantial performance differences between adaptive and CRD solutions (see figure 2), which translates into meaningful cost savings and deployment efficiencies. FIGURE 2: Outdoor testing, single client CDF, Single iphone 4, 2.4 GHz Chariot TCP Throughput, Mbps Typical CRD example Typical CRD example Ruckus 7762-T DL Ruckus 7762-T UL % of samples (n=40) with throughput >Y value Site acquisition is never cheap in outdoor environments where small cells are most needed. The AP s ability to effectively bear heavy client loads will directly influence capital and operating costs (not to mention user experience). Once again, substantial differences exist between adaptive and CRD performance, providing 2x (or more) CAPEX efficiency with adaptive systems (see figure 3). FIGURE 3: Effective density handling with adaptive systems. Source: Syracuse Center for Convergence and Emerging Networking Technologies (CCENT) High Density: 90 active clients per AP Interference adaptive antennas + capacity-based channel selection total TCP throughput conventional wi-fi implementations 95% improvement (average) Operating Interference: small-cell networks 6 APs, necessarily 120 clients, involves 1 busy interference, rogue AP in a couple of different forms. Whether licensed or unlicensed, networks of small cells need to be constructed to minimize selfinterference, from the many nodes that can see 65% each other in most public spaces (whether indoors or out). In improvement the case of licensed nodes, the potential mutual interference (average) with macro/ micro radios operating on the same carriers must be considered. In the unlicensed bands, there is obviously an additional and potentially show-stopping dynamic in terms of interference sources coming from outside the small-cell network. Because of the pervasive deployment of enterprise and residential Wi-Fi in most urban settings, it is possible for an outdoor network radio to see tens if not hundreds of individual WLAN networks occupying the same bands at the street level. Despite the low amount of user-plane traffic on these networks, excess control overhead (i.e. beaconing and scanning) can eat an excessive amount of channel capacity and it s sent at low modulation rates, making it receivable at great distances. This co-existence phenomenon can also influence indoor WLANs at single-room, public street-facing venues such as bars, restaurants, or convenience stores neighboring WLANs reach through front windows to pollute interior spaces. By contrast, this is often less of an issue in larger public venues such as train stations, shopping malls, and the like, because indoor penetration is limited by more substantial exterior construction and attenuated further by inner walls. page 3

4 n Nationally-televised two-day education summit, on Rockefeller Plaza in NY n 350 interfering APs nearby n Verizon, NBC s provider, was turned down by incumbent Wi-Fi vendors n Six Ruckus APs delivered 200 GB of traffic to hundreds of concurrent users n Education Nation guests and NBC staff raved Adaptation to Interference As an example of the scale of the problem and the difference that adaptive technology can make in this context, Ruckus recently deployed an event network for the Education Nation conference in Manhattan, held every year on the Rockefeller Center Plaza. With nearly saturated channels and still more users to accommodate, Ruckus moved an impressive amount of data, keeping users, operators, and staff all quite pleased. The principles of Ruckus interference management are three-fold: (1) Adaptive antennas the inherently better AP-to-client connections that the adaptive antenna approach allows can be used to extract more capacity from a given amount of airtime. At all ranges, adaptive antennas deliver higher SNR/SINR, delivering more efficient modulation and coding schemes and reducing errors/retries. This enables individual transmissions to be more efficient, and it simultaneously has a lower likelihood of corruption by interference, which is inherently sporadic and unpredictable. (2) Adaptive channel selection with active measurement by implementing active radio resource management techniques instead of passive options, Ruckus APs converge on the optimal channels with the highest capacity and least interference. Ruckus algorithms measure actual realized capacity, avoiding the specious guesswork innate to passive background scan techniques. FIGURE 5: Assessing Realizable Capacity: Visualization of ChannelFly s assessment of realizable capacity in 2.4 GHz channels. 2.4GHz RF CHANNEL FIGURE 4: Improving airtime efficiency and decreasing interference with adaptive radio links TIME (microseconds) OBSERVED THROUGHPUT (Mbps) page 4

5 FIGURE 6: ChannelFly convergence and ongoing operation: ChannelFly activity falls over time as channel characteristics in a new environment are learned FIGURE 7: Third-party ChannelFly testing: ChannelFly testing in network environments with busy neighboring interference sources confirms its value During the first minutes in an environment with or without clients ChannelFly gathers representative capacity data for each channel. After this learning period, the AP will change channels less frequently but continue to monitor each channel s capacity over time. Ongoing channel changes are made when capacity drops precipitously, while avoiding shifts for minor changes. Most competing channel selection approaches rely solely on background scanning to select a channel. APs leave their operating channel, passively listening to activity (retries, interference, frame count) on other channels for a very brief period of time. This approach suffers from a few key problems. First, passive metrics like retries and frame count are poorly correlated with actual channel capacity, decreasing predictive accuracy. Second, the AP abandons its operating channel for the duration of its background scans, leaving connected clients and their applications without packet flow. Finally, to minimize the second problem, APs must not leave their operating channel very frequently or for a long duration, which makes the off-channel data a poor representative sample of on-going channel activity. This technique simply does not provide enough of the right data for accurate channel selection. The net result is wasted capacity. Despite the many advantages of unlicensed frequencies, infinite capacity is not one. (3) Directional / narrow sector antennas a range of integrated directional antenna options, in both 30 and 120 sectors, allows the small cell coverage and capacity to be focused on the outdoor target area. This is a key feature in the outdoor urban environment where the objective is to add capacity along the busy streets and public gathering places. Typically, access nodes will need to be mounted to street furniture or the sides of buildings, and the ability to narrowly concentrate the beam down the street or to cover a sector of a ChannelFly was designed to maximize network throughput specifically in high-density, noisy public Wi Fi environments. Leveraging the same Ruckus-patented BeamFlex adaptation algorithms that learn and select the best signal path, ChannelFly is a statistical channel selection technique optimized to maximize airlink throughput. It relies on real-time observed performance on all channels in both 2.4GHz and 5 GHz frequencies to move clients to a better channel with less interference and higher capacity. By operating on all channels (not just non-overlapping ones), ChannelFly is far more effective than any self-generated, self-optimized plan based on manual tuning, as well as automated plans based on a limited choice of theoretically nonoverlapping frequencies (e.g. channels 1, 6, 11). FIGURE 8: Outdoor Demand Exists in Corridors page 5

6 FIGURE 9: High Indoor Demand and Utilization NLoS essential in tough locations Corners are obviously ideal but not always available city square is critical. Omni-directional antennas are often the wrong solution for the outdoor cityscape, as they will radiate much of the RF energy in sub-optimal patterns, decreasing overall network capacity. Deployment Location Challenges (Both LoS and NLOS) Deploying small cells in dense urban environments is challenged by pragmatic constraints at a level much more fundamental than the complex technology on which we tend to concentrate. Many challenges come from the simple problem of finding practical, affordable places to mount the radios. Mounting candidates such as light or power poles, street signals, billboards, bus stops, phone booths, building facades, and so on are typically governed by aesthetic, access, and connectivity (both power and signal) limitations, as well as complex business/ownership relationships needing navigation. The real-world compromise is to opt for mounting locations and configurations that are feasible (given all the above constraints) but not typically ideal in terms of getting the maximum incremental air interface capacity where it s required. In the face of these compromises, even with the simplest deployment requirements, the impact on small-cell network design can be significant. There are at least three fundamental considerations to make in relation to the limitations imposed by practical, physical sites. (1) Self-interference reduction one of the natural implications of sub-optimal radio locations is self-interference among multi-node deployments. Dynamic directional power steering with BeamFlex enables Ruckus to reduce self-interference (i.e contention) in multi-ap deployments by sending modulated signals only in the direction of the intended recipient. In multi-ap environments, this reduces adverse effects from unnecessary but completely normal and standardized protocol deferment and arbitration. 40 Mbps NLoS performance demonstrated in simple proof of concept with existing (nonoptimized) hardware. Product roadmap to 100 Mbps clear. FIGURE 10: Asymmetric BeamFlex patterns and interference mitigation: Directional power control reduces unnecessary contention with unintended recipients and neighboring cells (2) NLoS performance (smart meshing) resulting from the limitations of deployment locations, NLoS performance can easily become a major limiting factor for a deployment, both for client connectivity and backhaul. Commonly, locations that are agreeable for equipment mounting are not readily suited to deliver wired backhaul. As a byproduct, mesh links become a lynchpin conduit for user traffic, demanding the best possible performance. Concerns for both link stability and capacity suddenly stand in the way of effective deployments. For client connectivity, the same fundamental stability and throughput conditions potentially prevent association or adequate user experience. page 6

7 Thankfully, the benefits of both BeamFlex and ChannelFly combine to provide additional link budget and capacity to absorb suboptimal NLoS conditions and maintain satisfying performance. Adaptive software and degrees of freedom in the RF system enable Ruckus APs to focus not only on theoretical best practices and heuristics that work well in a lab environment, but more effectively to adapt to what actually works in a live deployment. Dynamic system adaptations can be determined by the results of first-hand performance metrics from each unique environment. This live adaptation is critical for NLoS environments, because manual optimization simply cannot account for the dynamic, characteristics of each venue; specifically, engineers cannot predict multipath, scattering, absorption and fading characteristics. Software-based learning and adaptation are the right answer. (3) Packaging efficiency driven by aesthetic and loading limitations on mounting assets, hardware design efficiency quickly becomes relevant to network operators. With many generations of custom antenna design and improvement, Ruckus RF technology fits neatly into attractive packages that lead the industry in both size and weight, as well as ease of installation. Marry the judicious design with best-in-class performance, and there s no need to sacrifice on optimal mounting locations or performance. Changing Conditions Outdoor radio environments are dynamic. They change all the time. In areas served well by small cells, several simple factors contribute: Street-level vehicle and traffic changes, especially taller buses and trucks, cause significant shifts in RF behavior Seasonally, foliage comes and goes in many regions User traffic (demand) profiles vary from moment to moment, with large swings across the span of a day, and potentially larger swings across seasons and events In the face of these changes, adaptive feature sets like BeamFlex, ChannelFly, and SmartMesh automatically find the best sets of operating characteristics to maintain optimal performance. Once again, these enhancements have twofold advantages by enabling more capable and more reliable backhaul as well as improving the initial access network directly to client devices. By definition, these environmental changes are an ongoing challenge, increasing operating expenses, maintenance, and system tuning when limited system adaptation is supported. Conversely, adaptive, smart Wi-Fi, has repeatedly proven to decrease the ongoing operating costs and time investment by operator staff. Customers with extensive outdoor experience are realizing the benefits, observing the access points making the intelligent choice for us, taking that pressure off of us 1 (Chris Spencer, CTO, Global Reach). Radio Integration Challenges Given the tremendously demanding nature of small cell applications, best-in-class solutions are a necessity. Wed that need with the significant constraints related to mounting assets (as previously mentioned), and it s no surprise that leading product suppliers and operators have increased interest in the integration, or co-location, of Wi-Fi with licensed-band small cells. Ruckus outdoor APs provide both PoE and downstream Ethernet 80% Less technical staff time Conventional Alternative Ruckus Average results from sample of 40 Ruckus case studies ~80% reduction in technical staff time required for WLAN installation & maintenance Maintain Troubleshoot Deploy connectivity to co-located nodes, enabling co-deployment of standalone licensed small cells with power and backhaul provided directly by the AP. As small cell deployments proliferate, for aesthetic reasons, municipalities will want to avoid over-occupied mounting assets, 1 html#prettyphoto page 7

8 Conclusion Building high-capacity RANs in urban settings is complex. Whether for technical, business, financial, geographical, or social reasons, there can be any number of limiting factors and challenges to effective implementations. Outdoor networks represent a perfect storm of opposition. High user densities, small and design-constrained client devices (low power, small antennas, limited battery), and interference all combine to make high-capacity connectivity more difficult. Pair those demands with highly dynamic and unpredictable RF environments, inhibited deployment locations, and LoS and NLoS complications, and suddenly off-the-shelf conventional Wi-Fi is unqualified. which begin to look like industrial coconut trees. Pressure has already increased along this trend. Naturally, to support multiple operators, street assets will be subject to the same one-networksupporting-all-operators (i.e. neutral host ) rule of deployment that emerged in the in-building cellular (i.e. DAS) market many years ago. The solution is a modular approach that combines best-inclass Wi-Fi (e.g. the SC8800) with licensed-band small-cell equipment at implementation time. Modular radio design enables easy neutral host through interleaved deployment of multiple operators licensed gear, by infrastructure, fixed-line, or wholesale operators. With this approach, small-cell licensed and unlicensed solutions can be planned in coordination with macro deployments. With any or all of these limitations to consider, smart solutions with adaptive software and multiple degrees of freedom in hardware will impact both initial deployment time and effort, as well as ongoing operations. Ultimately, several key enhancements are necessary: Adaptive antenna optimization with directional power control Adaptive polarization diversity Adaptive channel selection Smart, adaptive mesh networking Integration with macro and licensed small-cell networks Tight packaging and thoughtful mounting hardware In combination with many other system optimizations, dynamic features are the right answers to the most pressing questions for small cell, high-demand, outdoor networks in need of solid ROI. By virtue of more data capacity, more consistent performance, and more dependable connectivity, real-world side-by-side testing will prove (and has already proven) this out. Adaptive Wi-Fi is the answer. Don t hit the streets without it. Copyright 2013, Ruckus Wireless, Inc. All rights reserved. Ruckus Wireless and Ruckus Wireless design are registered in the U.S. Patent and Trademark Office. Ruckus Wireless, the Ruckus Wireless logo, BeamFlex, ZoneFlex, MediaFlex, FlexMaster, ZoneDirector, SpeedFlex, SmartCast, SmartCell, ChannelFly and Dynamic PSK are trademarks of Ruckus Wireless, Inc. in the United States and other countries. All other trademarks mentioned in this document or website are the property of their respective owners. Revised May Ruckus Wireless, Inc. 350 West Java Drive Sunnyvale, CA USA (650) Ph \ (408) Fx