Picocell Mesh: Bringing Low-Cost Coverage,

Capacity and Symmetry to Mobile WiMAX

A Business Case White Paper
March, 2007

Photo courtesy of NASA Image eXchange.

Image use in no way implies endorsement by NASA of any of the products, services, or materials offered by Tropos Networks, Inc.
A Tropos Networks White Paper
Picocell Mesh: Bringing Low-Cost Coverage, Capacity and Symmetry to Mobile WiMAX

Executive Summary
Mobile network user requirements continue to expand with the rapid emergence of
mobile Internet devices and Web 2.0 applications. The proliferation of these devices and
applications are fundamentally reshaping the requirements for mobile broadband wireless
infrastructures. This shift is driving the demand for networks that offer uniform coverage,
increased capacity and symmetrical client links so that users can experience at least 1 Mbps
uplink and downlink service wherever they require it and regardless of usage intensity.
Accomplishing this goal requires deployment of dense, low-cost, small meshed picocells
(typically mounted on streetlights), regardless of radio technology used.
Integrating mobile WiMAX into Tropos® MetroMesh™ systems offers an easy and cost-
effective path to providing the coverage, capacity and symmetry necessary to take Web
2.0 on the go. Wi-Fi/WiMAX MetroMesh systems also offer the flexibility of using either
licensed or unlicensed spectrum in the mesh network and can generate significant revenue
today from Wi-Fi clients while laying the foundation for future applications served via
licensed band mobile WiMAX. Finally, building out Wi-Fi MetroMesh systems today
and adding mobile WiMAX tomorrow can secure what will undoubtedly be increasingly
valuable mounting assets, such as streetlights, with no net cash outlay.
This paper starts with a review the requirements of Web 2.0 and its mobile users. It then
explains why traditional “four cells per square mile” macrocellular networks cannot meet
these requirements and why a system of low-cost picocells deployed at 20-40 cells per
square mile is the only way to deliver the coverage, capacity and symmetry that new
applications require. We close by explaining why sophisticated mesh routing software such
as Tropos MetroMesh NG™ is required to keep backhaul and management costs to
a minimum in these picocell networks.

The Internet is Going Mobile

The Mobile Internet offers a variety of applications and services to a plethora of device
types. A common theme underlies this diversity – the evolution from the 1990s model of
Internet access (Web 1.0) to the new model of Internet interaction (Web 2.0). An estimated
104 million Wi-Fi client devices will ship in 2007, bringing the installed base of Wi-Fi
clients to nearly 280 million by the end of the year1. These device types cross the full
spectrum from more traditional consumer and business mobile devices such as laptops,
PDAs, multi-mode handsets to newly Wi-Fi equipped cameras, MP3 players, gaming
systems and other consumer electronic devices. The applications range from Internet
____________________

1
2007 shipments from Xirrus. Installed base is a Tropos estimate based on Xirrus and IDC data.

2
A Tropos Networks White Paper
Picocell Mesh: Bringing Low-Cost Coverage, Capacity and Symmetry to Mobile WiMAX

interaction for mobile municipal workers to device telemetry for applications such as
utility meter reading, wireless parking meters and IP video cameras for surveillance in
high-crime areas. A variety of location-based services involving context-sensitive
advertising and content delivery are emerging. In an increasingly mobile world, Internet
users are demanding any place, any time, any device mobile broadband access.

The Mobile Internet (Web 2.0) Drives New Requirements

Mobile Internet applications and devices have significantly different needs than Web 1.0
applications. In particular:

• Symmetrical links are needed. The asymmetric “broadband” that is provided by today’s
3G cellular networks typically delivers less than 100 kbps on the uplink. This limited
uplink cannot practically accommodate the new trends of interactivity and user content
generation (video uploads, gaming, large file sharing, e-mails with large attachments).
Mobile broadband networks capable of providing high-speed uplinks as well as downlinks
are becoming mandatory. The small size and limited battery life of mobile devices
generally constrain their transmit power to 23dBm or less, which further exacerbates the
symmetry challenge.
• Uniform coverage is needed. Users expect consistent service and experience, regardless
of where they are in the network. Picocell-based metro-scale Wi-Fi networks today are
delivering symmetrical 1+ Mbps to over 90% of end-users. Our experience in deploying
and operating these networks is that customers simply expect this level of performance
in their broadband networks, regardless of whether they are wired or wireless. What
matters most is not median throughput (uplink and downlink) but the throughput that 90+
% of users can consistently obtain. When it is not true broadband, satisfaction suffers
significantly.
• Indoor coverage is needed. The current generation of mobile devices is increasingly
Wi-Fi-equipped. We expect that the next generation of these devices will be dual-mode,
with WiMAX capabilities built-in. While Wi-Fi suffers from challenges with in-building
penetration, consumer expectations are that WiMAX, with its superior reach and link
budget, will be able to provide significantly improved indoor coverage. We discuss
below the challenges to delivering on this promise in the context of traditional macrocell
WiMAX architectures and how a picocellular approach solves this problem.

These requirements, taken together, demand significantly higher cell density than today’s
macrocell cellular networks provide, as discussed below.

3
A Tropos Networks White Paper
Picocell Mesh: Bringing Low-Cost Coverage, Capacity and Symmetry to Mobile WiMAX

Wi-Fi Paves the Way to WiMAX

In addition to new applications and devices emerging, the number and diversity of Wi-
Fi devices continues to grow at an unprecedented pace. With 802.16e chipsets becoming
available now, mobile WiMAX devices will begin to appear in material numbers within the
next one to two years. The dense cells required to support mobile WiMAX can also be used
to deliver profitable services to Wi-Fi devices today, while providing a seamless migration
path to dual-mode Wi-Fi and WiMAX access for the next generation of mobile devices.
Finally, by deploying meshed Wi-Fi, mounting assets critical for picocell WiMAX, such as
streetlights, can be acquired today with no net cash outlay.

Today’s WiMAX Macrocells Will Not Support the Mobile

Internet
The macrocellular approach being pursued today by WiMAX vendors and carriers is similar
to the one that has been conventionally used for cellular telephone networks. While this
deployment model meets the needs of voice, which requires about 8 kbps ubiquitously, it
falls short of the requirements of Web 2.0 broadband data. Approaches today target one
to four base stations per square mile, with most deployments tending towards one base
station per square mile. Based on our experience building Wi-Fi networks for low-power
mobile devices as well as extensive modeling WiMAX propagation and performance,
this significantly underestimates the node densities required to achieve necessary uniform
high-bandwidth coverage to mobile clients. This, in turn means that subscribers will be
dissatisfied as they adopt Web 2.0-style mobile applications and devices.
Currently planned deployments also target using 30+ MHz of licensed spectrum in one
of several frequency bands (2.3 GHz, 2.5 GHz, 3.5 GHz, etc.), where propagation is
particularly challenging. As an attempt to compensate for poor performance at the cell edge,
operators are targeting a three-sector deployment of WiMAX where adjacent sectors operate
on different channels. This requires at least 30MHz of spectrum, which is not always
available at lower frequency bands. As regulatory bodies become increasingly aware of the
need to offer more wireless bandwidth, they are contemplating licensing wide swathes of
spectrum in higher frequency bands (2.5GHz and higher), mainly because of availability.
The World Radio Conference 2007 (WRC-07) is expected to harmonize spectrum for IMT-
advanced technologies in these higher frequency bands. These spectrum bands inherently
need small cells.
The macrocell deployment of WiMAX is particularly limiting for true broadband data on
the uplink. Since mobile clients are constrained in terms of transmit power, the receive
power at a base station in a macrocell scenario can be quite low. Our simulations show

4
A Tropos Networks White Paper
Picocell Mesh: Bringing Low-Cost Coverage, Capacity and Symmetry to Mobile WiMAX

that at densities of four base stations per square mile, in excess of 75% of users achieve
less than 1 Mbps of uplink throughput. This is inconsistent with the requirements of Web
2.0 applications such as video uploads or file sharing. Moreover, the available capacity is
not uniformly distributed – users close to the base station get high throughput, but users
further away get much less throughput. This can result in an inconsistent user experience,
especially in light of mobility requirements. We will quantify this in the next section of this
white paper.
The macrocellular approach doesn’t offer any easy way to scale the capacity in line with
growth in usage, since there are a finite (and small) number of high points (such as towers
or tall buildings) per square mile. Upgrading macrocell radios to the next generation
technology can offer some relief. However this will not address the primary issue of low
client transmit power.
Another limitation of the macrocellular approach is the poor indoor penetration offered.
Penetrating one or two walls can easily result in a link budget loss of 10-20 dB, rendering
the service unusable. We’ve encountered the same effect in our Wi-Fi deployments and are
moving, as an industry, to tackle it through higher cell densities as well as higher power
client devices.
Finally, greenfield WiMAX deployments miss the opportunity to reach the huge installed
base of Wi-Fi clients. With an installed based of nearly 280 million Wi-Fi client expected
by the end of 2007, Wi-Fi is approaching ubiquity in terms of penetration into all kinds
of consumer electronic devices. WiMAX may begin to achieve a similar status in the
2009-2010 timeframe. In the interim, providing access to the installed base of Wi-Fi
clients presents significant near-term revenue opportunities and secures subscribers.
When WiMAX clients appear, Wi-Fi and WiMAX will coexist, driven by the increasing
availability of dual-mode Wi-Fi/WiMAX devices. These networks can over time easily
migrate to dual mode Wi-Fi/WiMAX access for maximum subscriber satisfaction and
revenue potential.

Small, Low Cost Picocells Provide Needed Coverage,

Capacity and Symmetry
Shannon’s Law states that the capacity of a wireless link is directly related to the signal-to-
interference/noise ratio (SINR) at the receiver. (Note that SINR is often represented as SNR,
with the implicit understanding that interference is captured in the noise term.) The SINR
ratio improves when the strength of the desired signal increases, or when the interference
decreases. (Thermal noise, the second component of the denominator term of the SINR,
is usually a receiver attribute and is not related to deployment.) Since mobile devices are
low power transmitters, achieving sufficiently high SINRs at the base stations to sustain

5
A Tropos Networks White Paper
Picocell Mesh: Bringing Low-Cost Coverage, Capacity and Symmetry to Mobile WiMAX

broadband speeds > 1Mbps on the uplink can be most easily achieved using a picocell
architecture. Picocell architectures result in improved signal strength as well as reduced
interference, and thus improve SINR markedly. Below we will quantify the throughput and
capacity gains achieved through a picocellular approach comprising many, smaller cells.
Let us first discuss the improvement in signal strength and the resultant coverage and link
budget improvements. This is most evident on the uplink, where limited transmit power at
the wireless client device often results in poor link budget, especially for broadband data rates.
Our simulations show dramatic improvements in uplink performance in a picocell
architecture. Whereas only 25% of users in a typical macrocellular deployment see
throughputs greater than 1Mbps on the uplink, more than 90% of the users in a picocell
deployment see throughputs greater than 1Mbps on the uplink. The Mobile Internet
requirement of 1Mbps or greater on the uplink for over 90% of users is achievable only
using a picocell approach, given current radio technology. As is evident from the chart
below, there is an improvement of 15 dB in the median SINR on the uplink from the move
to picocells.

Figure 1: Improvement in SINR on the uplink from picocells

While cellular downlinks do not often suffer from chronic coverage problems, picocell
deployments result in significant gains in downlink performance as well. This is the result
of more efficient last hop links to the client that are able to operate at higher data rates and/
or lower power, leveraging proximity to the client device. A direct consequence of a dense
deployment is greater uniformity in the signal strengths, and consequently data rates that
are achievable across the network.
A second major benefit of picocell deployments is the reduction in interference. This is a
subtle point, and is often overlooked in the classical cellular macrocell paradigm. In a given
geographical area, there are a typical number of wireless clients that are concurrently active.

6
A Tropos Networks White Paper
Picocell Mesh: Bringing Low-Cost Coverage, Capacity and Symmetry to Mobile WiMAX

This number depends on population density and will increase with time as wireless usage
and applications become widespread. First, consider a classical macrocell deployment. At
even moderate population density, the duty cycle in a macrocell network is high, because
the macro base station must cover a large area with many concurrent clients.
In a picocell deployment, the number of concurrently active clients has not changed, but
the number of infrastructure elements has increased substantially by an order of magnitude
or more. Therefore, only a small fraction of these infrastructure elements are concurrently
active. From the perspective of a given picocell, many of its neighboring picocells are
statistically unlikely to be active on either the uplink or the downlink. This results in
improved SINR since the sources of interference tend to be scattered further away while the
signal strength to/from the client has increased due to proximity.

Figure 2: Improvement in downlink SINR as a result of reduced duty cycle

Figure 2 illustrates the improvement in downlink SINR that can be realized in a picocell
deployment as compared to a conventional macrocell deployment. The latter scenario
is described by the curve on the left that is labeled “4 nodes, 100%”, which implies that
there are four macrocells per square mile that are operating with a 100% duty cycle. In the
picocell scenario, as the number of infrastructure elements increases to, say, 40, we assume
that the duty cycle will reduce to about 10%. This is essentially a linear reduction in duty
cycle of an order of magnitude that is proportional to the order of magnitude increase in the
number of infrastructure elements in a given area. A gain of about 6 dB in median SINR is
realized as a direct consequence of the interference being statistically distributed across the
array of picocells rather than being constantly generated by neighboring cells.

7
A Tropos Networks White Paper
Picocell Mesh: Bringing Low-Cost Coverage, Capacity and Symmetry to Mobile WiMAX

Further duty cycle reductions are achievable beyond what is modeled here, since the
increased proximity of picocells to the end-user results in fewer airtime resources (time and
power) being required to transport a bit to or from the end-user.
This increase in SINR has a very dramatic effect on throughput as captured in Figure 3.
The consistency of the user experience is underscored by the fact that 90% of the clients
can experience 2 Mbps+ on the downlink in a picocell deployment with 40 nodes and a per
node duty cycle of 10%. In a fully loaded macrocell deployment with four base stations per
square mile, only about 60% of clients are in a position to experience these data rates.

Figure 3: Distribution of client throughputs as a function of node density and duty cycle

These SINR improvements and duty cycle reductions directly translate to increases in the
concurrent subscriber capacity (Mbps/square mile): a concurrent subscriber capacity of 40
Mbps per square mile for a single-sector macrocell deployment with four base stations per
square mile compares to a concurrent subscriber capacity of 64 Mbps per square mile for a
picocell deployment with an equivalent number (four per square mile) of capacity injection
points. Additional capacity gains can be achieved through the addition of capacity injection
points (mesh gateways) where feasible (based on line-of-sight to traditional macrocell base
stations).

8
A Tropos Networks White Paper
Picocell Mesh: Bringing Low-Cost Coverage, Capacity and Symmetry to Mobile WiMAX

To summarize, a picocell implementation can provide the following benefits:

• Better coverage
– More uniform coverage (fewer dead zones)
– Improved indoor coverage (enables addressing a larger subscriber base)
– True broadband with symmetric throughput (supporting Mobile Internet requirements)
– Higher data rates at the cell edge (allowing operators to offer higher SLAs)
• Higher capacity
– More evenly distributed (allowing operators to offer higher SLAs)
– Optimum use of licensed spectrum resources (efficient use of scarce licensed spectrum
resources)
– Efficient use of Wi-Fi frequencies (allowing operators to leverage ~500 MHz of free
spectrum where appropriate)
– Optimized client links operating at high data rates (higher end-user satisfaction)
– Scales capacity with usage by adding picocells (upgrade path using today’s
technology)

These benefits are achieved through

• Significant improvements in SINR on client uplink

– Median SINR improvement of 15 dB versus macrocell deployment
– > 90% of users experience 1+ Mbps on the uplink (versus 25% of users in a macrocell
deployment)

9
A Tropos Networks White Paper
Picocell Mesh: Bringing Low-Cost Coverage, Capacity and Symmetry to Mobile WiMAX

• Significant improvements in SINR on downlink to the client

– Median SINR improvement of 6dB dB vs macrocell deployment
– 90% of users experience 2+ Mbps on downlink (versus 40% of users in a macrocell
deployment)

This result should not be surprising. The system capacity of wireless networks has doubled
approximately every 30 months and a significant portion of the capacity increases can
be attributed to more efficient spectrum re-use through the employment of smaller cells
(Cooper’s Law, due to Marty Cooper). While capacity gains in wireless systems can be
realized through a variety of techniques including expanded use of spectrum, improved
modulation techniques, advanced signal processing techniques such as MIMO, etc., for a
given radio (MAC/PHY) technology implementation, the most effective way to increase
overall system capacity is to employ efficient frequency re-use through the use of smaller
cell sizes.

Tropos MetroMesh NG: The World’s Only Multi Radio

Protocol Picocell Mesh System
Picocells, while attractive, must be much lower cost than today’s macrocell base stations.
They must also have a smaller footprint and consume much less power to achieve a
form-factor suitable for mounting on streetlights. They require a mesh architecture, since
it is infeasible and uneconomical to run backhaul to a large number of cells per square
mile. Management of the large number of cells required to effectively serve even a small
city becomes extremely problematic without automated, distributed intelligence and the
collection and analysis of large amounts of management data.
Furthermore, since spectrum and airtime are finite resources, intelligence is needed to
enforce policies at the network edge and consistently extend them to the forwarding
plane within the mesh. Edge intelligence is also needed to achieve fast Layer 3 mobility
that tracks the Layer 2 handoffs and efficiently and quickly sets up routes. In general,
complexity and cost can rise exponentially as cell sizes shrink and distributed software
intelligence at the edge is a requirement in order to meet the goals of eliminating
complexity and reducing cost.
There are many challenges to be addressed in developing a picocell mesh solution –
challenges that Tropos has encountered while deploying MetroMesh networks in the
considerably more difficult arena of unlicensed band Wi-Fi deployments. The MetroMesh
architecture with the Predictive Wireless Routing Protocol (PWRP®) was developed to
address these challenges and has gained the benefit of widespread deployment. Tropos

10
A Tropos Networks White Paper
Picocell Mesh: Bringing Low-Cost Coverage, Capacity and Symmetry to Mobile WiMAX

Wi-Fi mesh networks, at densities of upwards of 30 routers per square mile, are delivering
Mobile Internet applications in more than 500 deployments today including ubiquitous Wi-
Fi access in such cities as Mountain View, Oklahoma City, to name a few. These networks
are being used to deliver symmetrical 1+ Mbps broadband outdoors to over 90% of users.
Below, we outline the challenges to developing a scalable, high-performance picocell mesh
system and describe how Tropos’ MetroMesh architecture with PWRP overcomes them:

• Scalable routing. Picocellular networks need to be able to scale to deployments of

thousands of nodes covering hundreds of square miles while maintaining low overhead.
This presents a challenge to traditional approaches such as distance vector and link state
routing protocols where the overheads grow linearly or quadratically with the size of the
network. PWRP is a scalable routing protocol, designed to scale to very large deployments
while keeping the routing overhead to < 5% of airtime utilization. The MetroMesh
architecture gracefully accommodates network expansion and capacity addition.
• Throughput-optimized routing. Mesh networks require a routing protocol that selects
the highest throughput and most reliable paths. Links with high packet error rates result
in high numbers of MAC-layer retransmissions and lower throughput. The routing
protocol needs to be aware of error rates and select the routing paths that minimize errors
and retransmissions. In deployments with 30+ nodes per square mile, the average node
often can choose from more than 10 available paths. The difference between choosing
the best path and any other path can often result in dramatically different throughput and
reliability. PWRP selects the highest throughput and most reliable paths through the mesh,
by design. Other approaches attempt to minimize hop count or examine next-hop SINR
and do not factor in end-to-end packet error rates, resulting in lower throughput.
• RF spectrum management. One key to achieving the dramatic duty cycle reductions is
the development of efficient algorithms for RF spectrum management. These algorithms
are required to make efficient use of spectrum through optimal spatial frequency re-use,
to provide dynamic interference rejection, to enable maximization of capacity, to enable
efficient concurrent operation in multiple frequency channels in a given area, to select the
optimum data rates for link operation and to tune MAC-layer packet detection registers
to adapt to the RF environment. PWRP continuously and dynamically optimizes the use
of available spectrum through the use of patent-pending algorithms for automatic channel
selection, adaptive data rate selection, automatic transmit power adjustment and adaptive
noise immunity (ANI). These algorithms have been extensively tested and verified in
several real world deployments and are growing extremely important as deployments
become denser and coverage requirements increase.

• Multi-mode routing. The routing protocol must incorporate the intelligence and the

11
A Tropos Networks White Paper
Picocell Mesh: Bringing Low-Cost Coverage, Capacity and Symmetry to Mobile WiMAX

flexibility to accommodate diverse operator-specified policies regarding preferences for

utilizing a combination of licensed and unlicensed frequencies and a variety of radio
protocols. It needs to intelligently route traffic over a mix of frequency bands to maximize
end-user throughput. On multi-mode platforms, PWRP has the ability to automatically
determine the optimal frequency band over which to close mesh links. This enables
the mesh to increase overall system capacity through the opportunistic use of multiple
frequency bands. In addition, the use of the higher capacity multi-mode product enables
larger clusters with higher overall hop counts without sacrificing throughput or capacity.
Tropos has numerous key patents on multi-radio mesh architectures dating back to 2001.
• Dynamic rate limiting, traffic management and IP services. Tropos has built and
deployed hundreds of multi-use networks designed to support a variety of user groups
and applications with differing requirements along the dimensions of security, QoS and
other policies. These policies need to be applied at the mesh edge and consistently applied
within the mesh, to manage the impact on bandwidth. As one example, in networks with
heavy usage, a small number of users often end up consuming a disproportionate share
of the bandwidth through applications like peer-to-peer file transfers. These high traffic
streams can have a negative impact on the network performance of the remainder of the
users, unless proactive traffic shaping policies are applied at the edge to these bandwidth-
hungry applications. Tropos’ Mesh Edge Service Management (MESM™) feature set
provides, among other things, the ability to proactively identify these classes of users
(based on traffic stream characteristics) and to apply selective rate limiting policies so
as to allow for a more equitable distribution of bandwidth and to apply more granular
operator-specified policies on network usage. This wide feature set was developed based
on operator feedback and an analysis of usage data gathered from some of our most
heavily used production networks.
• QoS. While MAC/PHY protocols offer some degree of QoS at the link layer, in a mesh
network these need to be integrated with higher layer routing and QoS management
functions. Tropos’ QoS implementation is designed to ensure that prioritization and QoS
policies can be applied at the mesh edge and consistently extended to the routing and
forwarding plane within the mesh.
• Mesh data collection. One of the key operational requirements for being able to deploy
and manage large-scale mesh networks is the ability to collect and correlate mesh-wide
usage and performance data across a very large number of nodes across the network, since
picocell networks can easily run into thousands of nodes per city. Another key requirement
is to be able to accomplish this critical data collection while keeping the airtime overheads
to a minimum. Tropos’ Correlated Mesh Data Protocol (CMDP™) enables the very
efficient collection and correlation of mesh-wide usage and performance data. Using this
protocol, the mesh provides precise information correlated across nodes on the network
about end-user usage, client device link quality, throughput across backhaul links and
mesh paths, mesh packet success probability, interference and airtime utilization. For

12
A Tropos Networks White Paper
Picocell Mesh: Bringing Low-Cost Coverage, Capacity and Symmetry to Mobile WiMAX

example, data collected using CMDP was used recently to optimize a production network
by generating automated recommendations on optimal nodes to convert to gateways in
order to improve overall system capacity.
• Compensating for Client Variability: Tropos developed the Adaptive Mesh Connectivity
Engine (AMCE™) to compensate for the considerable variability exhibited by mobile
devices in terms of output power, rate and power control algorithms, etc., to ensure a
uniform and high-quality user experience across device classes.
Tropos’ MetroMesh NG™ adapts the core radio-agnostic MetroMesh architecture to mixed
Wi-Fi/WiMAX and provides support for both Wi-Fi and WiMAX clients using picocells.
MetroMesh NG seamlessly integrates into deployed macrocell WiMAX networks, allowing
operators with existing WiMAX macrocell deployments to create a picocell underlay
network to boost capacities and throughputs and provide reliable coverage indoors and
outdoors to low-power mobile devices. The MetroMesh network can use the overlay
macrocell network for capacity injection at gateway points within the mesh with integrated
WiMAX subscriber stations. Meanwhile the picocell mesh ensures uniformly distributed
coverage and capacity through the operation of high-link-budget links to and from the client
device, whether it is Wi-Fi or WiMAX.
The radio-agnostic MetroMesh platform has been extended with MetroMesh NG to
incorporate Wi-Fi and WiMAX radios and Tropos’ Spectrum and Application Based
Routing Engine (SABRE™) intelligently selects wireless routing paths that maximize
throughput and capacity, cutting across frequency bands and radio standards.
SABRE can be programmed to run exclusively in licensed spectrum, or to use licensed
spectrum and unlicensed spectrum in combination to maximize operator capacity and return
on investment while maintaining required client SLAs. Especially on the client-facing
link, WiMAX in licensed spectrum offers significant advantages to operators in terms
of interference immunity and scalability. At the same time, there is a significant amount
(~500 MHz) of unlicensed spectrum in the 2.4 GHz and 5 GHz bands that it makes sense
to leverage as backhaul when it is available and interference-free. SABRE incorporates
intelligence to dynamically sense interference and adjust routing paths to avoid interference
and to select the routing path that achieves the best throughput and capacity, cutting across
frequency bands and radio standards. This allows for efficient utilization of unlicensed
spectrum where possible, thereby increasing overall capacity and protecting the scarcer
licensed spectrum resources. The Virtual Radio Interface, introduced as part of MetroMesh
NG, allows for the abstraction of the underlying radio technologies from the higher layer
protocols, including PWRP routing.

13
A Tropos Networks White Paper
Picocell Mesh: Bringing Low-Cost Coverage, Capacity and Symmetry to Mobile WiMAX

MetroMesh NG, which extends the MetroMesh architecture, incorporates all of the
MetroMesh software that powers 500+ deployments. Its picocell RF spectrum management
and IP routing, control and analysis features bring five years of field experience from
heavily-utilized metro Wi-Fi systems to mobile WiMAX. It incorporates the Tropos
Analysis and Control system including Tropos Insight 2.0 business analytics and CMDP for
processing and correlating mesh statistics to provide backhaul, mesh and client visibility.
MetroMesh NG also incorporates AMCE™ connectivity software to ensure reliable
network-based connectivity to full range of Wi-Fi devices and MESM IP services software
used to create secure, multi-use networks with QoS, mesh edge rate limiting and to offer
value-added services. Tropos’ PWRP routing and RF spectrum management algorithms
are included and extended through the use of SABRE routing technology that maximizes
capacity, fault tolerance and flexibility and BGP routing support for mobility at scale.

14
A Tropos Networks White Paper
Picocell Mesh: Bringing Low-Cost Coverage, Capacity and Symmetry to Mobile WiMAX

Summary—Mobile Internet Needs Small, Low Cost

Picocells; Picocells Need MetroMesh NG
The MetroMesh NG picocellular architecture ensures more uniform client coverage
including indoors. The improved coverage allows an operator to profitably increase their
subscriber base by expanding their addressable market. This enhanced coverage is achieved
while supporting high and symmetric multi-megabit data rates to low-power client devices.
Data rates and throughputs at the cell edge are improved dramatically as well.
The use of SABRE enhances system capacity by enabling efficient use of licensed spectrum
resources while allowing opportunistic use of unlicensed (free) Wi-Fi frequencies when
possible. This increases overall system capacity and ensures that the capacity is uniformly
distributed throughput the coverage area. Through the picocell architecture, client links
can operate at higher data rates, thereby maximizing throughput while minimizing airtime
utilization as well as interference. As the usage grows, operators can easily scale capacity
by adding more picocells to the system.
The MetroMesh NG approach can provide full support for Wi-Fi clients, allowing operators
to tap into this near-ubiquitous client base. It also supports WiMAX client access, at higher
data rates and with enhanced QoS. It thereby creates a smooth migration path from Wi-Fi to
WiMAX and beyond (802.16m, etc.).
The key to achieving these benefits is the underlying software-driven low-cost picocell
architecture. MetroMesh OS software has been designed from the ground up to be radio-
agnostic and capable of operating in multiple frequency bands and across multiple radio
standards. The software and systems have been proven in over 500 deployments to date
and has achieved full interoperability with a wide diversity of client devices. The same
architecture also allows for a seamless integration into existing macrocell WiMAX
deployments as an underlay.
The support for uniform coverage at symmetrical broadband speeds to (and from) low-
power client devices is what is needed to fulfill the requirements for Web 2.0 applications.
With MetroMesh NG, reliable coverage extends both indoors and outdoors and at the high
speeds required for Mobile Internet applications.

15
A Tropos Networks White Paper
Picocell Mesh: Bringing Low-Cost Coverage, Capacity and Symmetry to Mobile WiMAX

Appendix: Simulation Assumptions

• Morphology:
– Macrocell scenario: 4 macrocell base stations per square mile
– Picocell scenario: 10 – 40 nodes per square mile, 4 gateways per square mile
– 400 clients per square mile
– Up to 2 network hops in the picocell mesh, typically 65% at depth-1, 35% at depth-2
– Typically 10 nodes per cluster (4 backhaul points per square mile in both scenarios)
• Propagation:
– 2.5 GHz spectrum band for WiMAX
– Path loss exponent of 3.5
– 28.6 + 3.5 * 10log(d) propagation model
• WiMAX system assumptions:
– 10 MHz – 30 MHz of spectrum, TDD, 2:1 DL:UL partitioning
– 0 – 25% time zones used for mesh depending on how much unlicensed spectrum is
brought into use.
– 10% overhead for MAP (control), 4% for preamble, guard, etc.
– Downlink: 2x2 MIMO (STC and SM) in addition to Receiver MRC.
– Uplink: 2-way MRC on the BS receiver, single transmit chain on the terminal
– Uplink interference floor assumed to be 5 dB over thermal noise.
• Duty cycle:
– 10 – 25% for picocells (100% for cluster gateways)
– 100% for macrocells

555 Del Rey Avenue • Sunnyvale, Ca 94085

phone 408.331.6800 • fax 408.331.6801
www.tropos.com • [email protected]
©2004-7 Tropos Networks, Inc. All rights reserved. Tropos and PWRP are registered trademarks of Tropos Networks, Inc. Tropos Networks, AMCE,
MetroMesh, TMCX, and Metro-Scale Mesh Networking Defined are trademarks of Tropos Networks, Inc. All other brand or product names
are trademarks or registered trademarks of their respective holder(s). Information contained herein is subject to change without notice.
The only warranties for Tropos products and services are set forth in the express warranty statements accompanying such products and services.
Nothing herein should be construed as constituting an additional warranty. Tropos shall not be liable for technical or editorial errors or omissions
contained herein, and any reliance by a party on any financial or other information contained herein shall be at the sole discretion of that party.

16