LocustWorld Open Source Mesh network are a
great wireless technology for indoor and outdoor use. But the most common mesh
architecture for the commercial market consists of routing packets over wireless
links to a wired network. This architecture is an excellent choice for wireless
ISPs (WISP) that want to create a wireless broadband network, such as 802.11b,
over a large geographic area. This paper is to assist the wireless Mesh Planner
to understand and develop their own capacity model or to help them choose a tool
or service that would assist them in their planning when leveraging the bands of
802.11b.
Mesh technology can deliver high bandwidth
at an order of magnitude lower cost than existing cellular technologies. As a
result, the cost of future mobile Internet access will be at a price point that
the average person can afford, giving way to a whole new market for wireless
devices and services, such as streaming video over handheld media players. In
business, the mesh architecture lets IT departments extend wireless coverage to
areas that do not have cabling infrastructure or require a mobile solution. In
these situations, MeshAP integrate with existing wired infrastructure to extend
Wi-Fi coverage to areas not readily accessible by cables.
Considerations
Designing wireless Mesh is a new craft, even
for many experienced network architects. When designing a wired network, most
network architects are familiar with the steps to ensure sufficient capacity for
the users and applications. With IEEE 802.11b, a new factor comes into play: the
tradeoff between radio-frequency (RF) coverage and bandwidth capacity per
device. Planning for both capacity and coverage is one of the key design issues
for Wireless Mesh. The key to your plan is for sufficient capacity, thus
adequate coverage will follow.
The biggest change in mindset when designing
Wireless Mesh is that they provide shared, not switched, connections. Many Mesh
Operators mistakenly focus only on providing adequate coverage, rather than
sufficient bandwidth, for their customers. While coverage may be the design
point if you are deploying a single point to point based network, the
application demands for a wireless ISP makes bandwidth the real design criteria.
You will not deliver enough bandwidth if you plan for coverage alone.
Key issues to consider when deploying a Mesh
include determining RF coverage from a single MeshAP, ensuring sufficient
capacity to support the population and accounting for RF signal-loss factors.
The distance at which a particular throughput can be achieved will vary
by the type of equipment that you are using, such that processor speed, radio
card, its sensitivity, antenna type and its power.
Selecting the operation mode of the MeshAP
can help in planning. In Ad-hoc mode many MeshAP will automatically decrease
their data rates as the RF signal degrades because a lower-frequency signal is
more likely to get through when there is interference. One user associated at 1
Mbps will slow the entire group, since the MeshAP will take longer to
communicate with that user, taking bandwidth from all other connected users.
Consider using Infrastructure mode rather than Ad-hoc mode.
It is important to point out, however, that
the addition of mesh access points can increase network latency. In an 802.11b
environment, each wireless hop created when a packet travels between a client
device and the wired network adds 1-20 milliseconds of delay.
As a result, careful design considerations need to be given to the size
of the mesh network and the types of applications that it is serving.
A major difference between designing for
wired and wireless LANs is the impact of objects on RF signals. Walls, doors,
windows and other fixed objects in the building will absorb RF signals, causing
signal loss. The building construction also has an impact: Concrete absorbs more
signal than wood. Look for tools that automate these calculations.
To avoid degrading the performance for all
users associated to an MeshAP, consider setting minimum association rates, in
Wiana Traffic Shaping section, that force users to associate with a new MeshAP
once their throughput falls below the minimum rate. By designing smaller cells
with higher throughput, you can create an enterprise-quality experience.
The number
of users and their applications are major drivers of bandwidth requirements. The
network architect must account for the number of users within the Mesh AP’s
cell diameter and the number of multi hops to the upload Meshap. In a large area
or where user density is high, you should design smaller cells to achieve a
higher data rate, since obstacles and objects will not naturally create the
cells by attenuating or blocking the RF signal. With smaller cells, you will
need to re-use frequencies more often and thus ensure that you are using
multiple channels.
Get the figures
Determining how much bandwidth each user
will need is critical, as your calculations will define the user experience, as
well as the number of MeshAP required. The
first part of planning your mesh model is to review the mesh architecture.
Better yet you should set up a small Mesh network with 3 nodes. An
uplink MeshAP, a repeater MeshAP and downlink Meshap.
Understanding the basic will help you plan your mesh network. The next
items are to have a plan for your network service.
It could be a business plan, a proposal etc. Having the prerequisite
information will help plan your network out.
Understand the technology and having will
help determine how many user or better yet how many client devices are you going
to support? There are many ways to
determine this number but a business plans specifically a marketing planning and
analysis with demographics will help determine the number of user.
Users or Client Devices = Number
of Devices (ND)
Capacity planning for your Wireless Mesh has
a critical issue to consider since wireless bandwidth is both shared and lower
than wired bandwidth. To understand
the amount of bandwidth you'll need, which determines the number and
transmission speeds of the MeshAP you’ll use, you need to characterize the
applications in question. How much bandwidth will they require to serve users
well? Will all the users in a given Mesh Cloud coverage area need access to the
same applications? If you don't already know the bandwidth (B) consumption for a
given set of applications, you can use a wired packet analyzer to provide some
application performance analysis. Another solution is to estimate the minimum
bandwidth (B) per user.
Once you know the bandwidth requirements for
a set of applications for ISP such as VoIP, HTML, Java, FTP, P2P, Multimedia
etc., multiply that bandwidth (B) by the number
of devices (ND) in an area to be covered by the Mesh.
Such that B= Bandwidth and ND=Devices,
B x ND
= Total Bandwidth (TB) of the Mesh Network
Now you'll have the total amount of
bandwidth that you need to service the Wireless Mesh. You may wish to reduce
this amount by an "over-subscription” factor, since it's unlikely that
all users will require the maximum bandwidth at the same time. Since end users
rarely require 100% of the available bandwidth all the time, ISPs will purchase
these expensive dedicated connections and use them to run multiple connections
of various sizes to their end users. This is called
"over-subscribing", and is a technique used by the Internet Service
Provider Industry to realize a profit margin when providing dedicated
connectivity to consumer users.
Over-subscription
(OS) is usually based on a bandwidth ratio and
typically ranges from 4:1 to 20:1 depending on the service being provided.
Over-subscription may actually occur several times before it reaches the end
user.
Over Subscription percentages:
Business = 5% – 25%
Co-Operatives = 25% - 40%
Community = 40% - 50%
Choose the percentage of Over Subscription
that best suites your business plan = OS%
Thus, Actual
Bandwidth to Service (AB) is Total Bandwidth (TB)
to service multiplied by the Over Subscription (OS)
factor, such that
TB x OS = AB
Wide Area Network
(WAN) connects through your router to your MESHAP.
In this scenario we need to determine the fast, most economical WAN
connection available at your location. You
may have many WAN connections to your Mesh; however, the assumption is that all
WAN bandwidth will be the same speed.
Wide Area Network
(WAN) available in your area = 1.5 Mbps
Once you know the expected total bandwidth,
you can calculate the number of Uplink MeshAP needed to provide that capacity.
Simply divide the Actual Bandwidth (AB) by the Wide
Area Network (WAN).
Actual Bandwidth
per MeshAP (AB)/Wide Area Network (WAN) = Number
of Uplink MeshAP (UM)
Because of several factors, actual
throughput on a wireless system is much lower than the technology’s specified
data rate. For instance, with even a one-way transmission on an 802.11b network,
the best possible Radio Card (RC) throughput is
4 Mbps to 6 Mbps.
For our model the Radio
Card (RC) throughput is assumed to be 5 Mbps.
Design note when implementing an aggregation
of uplinks nodes, (Aggregation is the combining of multiple remote WAN network
(i.e. ADSL, Cable, T1 etc) connections see Aggregation
using Mesh Nodes,) There is a higher availability of throughput compared to
using Non aggregation.
For example when adding the 10 Uplink node
to a small WAN connection, such as ADSL (1.5 Mbps), the maximum speed is limited
to the WAN connection and to the radio equipment.
This means the highest speed obtainable at the uplink nodes is WAN
connection speed and throughput; however, when using aggregation of the WAN
connection you are achieving a higher speed and throughput.
Having 10 ADSL WAN connections can give different result in an
aggregation and non aggregation setup, such that,
Divide that
MeshAP coverage area into the total coverage area to determine the number of
MeshAP needed for coverage, such that,
Service
Area (SA)/ (Benchmark
(BM)) squared – the total Uplink MeshAP Nodes
(UM) = Total number of MeshAP required.
We assume that the Uplink node will cover part of the Service Area, thus we
subtract that from the total covered area.
SA
/ BM2 - UM = Total Required MeshAP (TRM)
Therefore TRM is
the Total Required Repeater and Downlink MeshAP.
The TRM and UM will make the total wireless infrastructure.
Total
Wireless Infrastructure (TWI) = UM + TRM
Please
note that this is an estimate. To get further details I suggest you read Metropolitan
Mesh Installation. This will
help to define a number of parameters that are not assumed in this model.
As flexible as
the LocustWorld software is you must consider an important point before
determining the Total amount of MeshAP is the topology.
The topology that you choose can have a great impact on your capacity
planning. For example, channel
assignments can assist in providing additional coverage. You can place them
appropriately in an area and provide their channel assignments. When allocating
channels to the cells, be sure that adjacent cells use non-overlapping channels.
802.11b provides three non-overlapping channels.
This model assumes a single channel assignment; however, when dealing
with multi dwelling units (MDU) you may require overlapping floors with
different channels.
Conclusion
Network managers should demand
enterprise-quality design and management tools for their Wireless Mesh–the
same quality tools that they have for their wired networks. Wireless Mesh design
tools should assist network managers with the design parameters, including city
or town size and topology, obstacles, throughput per user, country of operation
and choice of 802.11 technologies. The tool should also automatically assess how
many MeshAP are needed, where they should be located and what their settings
should be.
This model makes
several simplifying assumptions. The worst assumption is not accounting for the
RF attenuation factors in your particular obstacle. Fortunately, RF capacity
planning is already available and made automatic in some Wireless system
planning management tools. The better planning tools assign these factors based
upon your geography and calculate the number, placement, and channel and power
settings for MeshAP based on your input of user bandwidth requirements. Another
alternative is to hire a firm to perform an RF site survey, but be sure they are
asking the right questions regarding your bandwidth requirements.
Build your model
of current usage per device, and then profile their locations. I have provided a
spreadsheet that contains a matrix
of information that is gather and a quick calculation to help determine your own
model. There are also packaged applications to help you plan your
bandwidth usage based on the number of users, number of access points, etc. As
you calculate bandwidth usage now, however, remember that your system should be
able to automatically adjust to changes dynamically. You should be able to move
users freely anywhere in the Mesh Cloud. They might get a lower megabyte per
second rate, but the user capacity should remain the same.
