Subnetting 101

More information on IP networking:

Most radio station networks that I have seen are divided along several different lines based on functions.  These functions are:

  • Office network; E-mail, document storage and retrieval, printing, applications like traffic and billing, promotions, music scheduling, and so on
  • Automation network; automation servers, workstations, and audio editing machines used in production
  • Audio over IP (AOIP) network; any AOIP consoles, devices, or STL equipment
  • Voice over IP (VOIP); telephone system
  • Wireless LAN; WLAN or WIFI

It is helpful, then, to segment the network into different broadcast domains to reduce the congestion on any one network.  That is where a good subnetting scheme can be beneficial.  Subnets segment the network into smaller parts, reducing the amount of broadcast traffic and increasing network speeds by reducing MAC table sizes, and thus switching and lookup times.  They also can secure certain areas of the network from the outside or other subnets, adding a level of security.  For example, it may not be a good idea for automation computers or AOIP consoles to have access to the internet.  Certain functions in routers and switches can be enabled for that added security.

It is also important to efficiently use IP addresses in a large organization where WANs are used.  The better the subnetting scheme, the easier it is to understand and the better it performs.   Avoiding or reducing discontiguous networks is key to efficient and speedy routing.   That is an important consideration where applications like AOIP and VOIP are concerned

To really understand subnetting, it must be broken down into fundamental parts.  This pertains to IPv4, which will likely remain in use for quite some time.  The big chart, class B networks:

3nd  octet4th octetCIDRDecimalWild cardHosts3rd Up bySubnets

Class C networks

3rd octet4th octetCIDRDecimalWild cardHosts4th Up bySubnetsBSubnetsC

The terms “Class B” and “Class C” networks are outdated.  Basically, I broke the chart up along a classful boundary to make it easier to read.

An IP v4 address consists of four octets of binary data. A common example is, which in binary numbers looks like this: 11000000.10101000.00000001.11111110. It is converted to base ten numbers (dotted decimal) so that we humans can deal with it. A typical subnet mask seen in many office networks is, which in binary looks like this: 11111111.11111111.11111111.00000000.  When a router receives a packet, it does something called an “ANDing process.”  When a router ANDs, it overlays the subnet mask on the network address and uses the following function: 1+1 = 1, 1+0 = 0 and 0+0 = 0.  Thus, in the above example, a router AND would look like this:

Dotted DecimalBinary Octets
192 168 1 254
255 255 255 0
192 168 1 0
11000000 10101000 00000001 11111110
11111111 11111111 11111111 00000000
11000000 10101000 00000001 00000000

The subnet mask is telling the router to ignore the last octet, thus saving a bit of time and processing power.  It may seem very small and insignificant.  When considering that routers make sometimes hundreds or thousands of routing decisions in a second, even a small bit of work reduction adds up quickly.  Subnet masks allow routers to look at only the layer three network address, ignoring the host portion.  This takes advantage of IPs inherent hierarchical addressing system and speeds the process of routing to the proper destination.

Another way to look at it:

IPv4 subnet chart
IPv4 subnet chart, click for .pdf version

There are three IPv4 address ranges set aside for private (internal) use:

  • to /16
  • to /12
  • to /8

Thus, very large networks can use an internal IP address scheme in the range and have up to 16,777,216 hosts, or 224 addresses minus two, one for the network line address and one for the broadcast address.  That would be one giant network clogged with ARP requests, ICMP packets and other miscellaneous multicast messages. A notation of /16 means that 16 bits are used for the network address, the remaining address bits are host bits.  A /24 network has 24 network bits and 8 host bits making the available hosts 254.

An example of an efficient network would be a medium market operation with six radio station under one roof.  This facility has ten studios and a newsroom using AOIP consoles, a VOIP phone system, an automation system, an office network with an internal file server and exchange server.  The number of required hosts on each subnetwork is

  • Office network, servers and wireless hosts: 78
  • VOIP phone system: 70
  • AOIP consoles and nodes: 30
  • Broadcast automation system: 22

Given IP address: /22

In most instances, office networks are usually installed on one class C segment, that is to say, the network mask is  However, in the example above, 254 hosts are not needed on the office network, thus it can be divided in half using the subnet mask of, leaving the other half for the VOIP phone system.  This subnetting scheme would leave 126 hosts on the office network and 126 hosts on the VOIP network.  The AOIP console and broadcast automation system can be placed on another class C segment, using the subnet mask of, which would give each subnet 62 hosts.  All subnets would have room to expand.  Each subnet is isolated from the others by a router.  The office subnet contains the gateway to the internet, usually .1 or .126 (first or last) IP address.

That would look something like this:

Office network
Line address First available Last available Broadcast Subnet mask
VOIP phone system
Line address First available Last available Broadcast Subnet mask
AOIP consoles and nodes
Line address First available Last available Broadcast Subnet mask
Broadcast Automation system
Line address First available Last available Broadcast Subnet mask

That keeps the network segments small but has room to grow.  This is a diagram of a converged network:

Radio Broadcast Facility converged network
Radio Broadcast Facility converged network

With a setup like this, reliability is the key to a happy life. The router should be a good Cisco product with four or more Fast Ethernet ports. A second way to do this would be to have four routers plugged into a distribution switch and use OSPF to route between subnetworks. The switches should also be a good Cisco product, which can take advantage of port security options and QoS on the VOIP and AOIP segments.  VOIP systems usually require Power over Ethernet (POE) ports, thus that switch can be specialized for that purpose.

Many AOIP systems want to see Gigabit switches or at least Fast Ethernet switches with Gigabit or better backplanes.  Any AOIP STL system can be connected to the AOIP network along with other things like AOIP remote broadcast and studio telephone solutions.

Many WLAN access points can be configured as a network router and DHCP server for wireless hosts.

The largest users of the public (i.e. internet) network would be the VOIP phone system and office network.  The broadcast automation network may also be a if voice tracking or other program delivery over WAN is used.

Nanobridge M5 wireless LAN link, Part II

After a bit of delay, we were able to return to the WICC transmitter site to install the Wireless LAN link.  The installation was pretty straightforward.  The studio unit was mounted on an existing STL tower on the top of the elevator room, the transmitter unit was mounted on an existing pipe on the roof of the transmitter building.

M5 Nanobridge mounted on transmitter building with radome
M5 Nanobridge mounted on transmitter building with RADOME

I included RADOMEs for a couple of reasons; first, there is a lot of critters around of the two-legged and winged kind. The upright two-legged critters may be attracted to the signal-strength lights at night. This unwanted attention could invite the juvenile delinquent’s bored teenagers to throw various objects found laying around on the ground at the antenna, damaging it.  The winged type critter may be inclined to view the feed horn as a good nesting location. The other reason is this site gets a lot of rain, wind, ice, and snow, therefore the RADOMEs afford some protection against the weather.

Aiming the antennas was pretty straightforward, but requires at least two people.  Using landmarks, we aligned the dishes in the general direction of each other.  Both ends of the system were turned on and we had a -89 dBm signal path, and somewhat surprisingly, the radios linked up and my laptop grabbed an IP address via DHCP.  Using the signal strength meter on the side of the antenna, each dish was peaked in turn:

M5 Nanobridge Antenna signal strength meter
M5 Nanobridge Antenna signal strength meter

Then, somebody on either end went below and looked at the signal strength screen on the web interface while the other end peaked.  In the end, we had about -65 dBm signal strength, which is somewhat less than the -58 dBm predicted.  I think we can do better, so on the next clear day, I am going to peak the signal again.

The data rate initially reported was over 100 MBPS, however, once I started transferring files back and forth, that dropped to about 50 MBPS.  If it is raining, that rate drops to about 35 MBPS, which is still far above what we need this link to do.  As a test, I streamed a youtube video, downloaded a Windows update, loaded several web pages, and checked my email simultaneously.  There were no issues with the data rate while those tasks were being preformed.

It is quite amazing to me that these little inexpensive radios can work so well.  My boss thinks that they will be blown up by lightning during the first thunderstorm of the season.  I don’t know.  There are several of these units have been installed at mountaintop tower sites and have been working for several years without issue.

Next step, installing the IP cameras and warning signs on the fence, setting up the monitoring software, etc.

Transmitter site security cameras
Transmitter site security cameras

Cameras mounted on old chimney platform.  This is the first set of cameras covering the south, north, and west approaches.  A fourth camera will be mounted on the back of the building covering the east approach.  Then, under the eves’ cameras will be mounted on all four corners of the building and the generator shed.  If anything moves, it will be recorded.

Ubiquiti Nanobridge M5 IP radio

I am in the process of installing a pair of the Nanobridge M5 units as an IP network link between a transmitter site and the studio location. The path is relatively short, about 1.5 miles over mostly water.  The main reason for this is to replace the analog phone lines used for remote control data and backup programming delivery to the transmitter site.  One added benefit, we are also installing several  IP cameras to keep an eye on the place.  We purchased the Nanobridge system for $80.00 per side.  The price is pretty good, but the configuration and testing are a bit intensive.

Network diagram
Network Diagram

There are many versions of these spread spectrum radios, some are licensed, and some are license free.  These are inexpensive, license-free links that I would count on for short paths or use in non-congested areas.  In congested areas, licensed (Part 101) links should be used, especially for critical infrastructure like STLs.

Since I dreamed up this idea, I figured I should make sure it is going to work before recommending it to the powers that be.  I have learned the hard way, almost nothing is worse than a failed project with your name on it.  Better to over-study something than to go off half-cocked, spend a bunch of money, then realize the idea was flawed from the start.  See also: Success has a thousand mothers but failure is an orphan.

Nanobrige path study, 5.8 GHz
Nanobrige path study, 5.8 GHz, moderate noise floor, 1.5 miles

Looks pretty good.  300 MB/s bi-directional which is faster than the Ethernet port on the unit.  This will be set up in bridge mode with pretty robust encryption.  The transmitter site side is configured in the router mode, creating a second class A network at the remote site.

Nanobridge M5 22 dBi antenna
Nanobridge M5 22 dBi antenna

Next step, configuring the units.  The Nanobridge units were set up in a back to back configuration in the engineering room.  Each end comes with a default IP address of  The units were several steps behind the latest firmware version, therefore the firmware was upgraded first.  The default admin user, password, and IP addresses were changed.  There is no greater security risk than default user and password.  The wireless security feature is enabled using WPA2-AES PSK and a greater than 192-bit access code.  The unit allows for any access code length up to 256 bits.  With a key of between 192 and 256 bits, the number of possible solutions is between 6.2771 E 57 and 1.1579 E 77, which should be pretty hard to crack.  By way of reference, a 192-bit password has 24 ASCII characters and a 256-bit password has 32 ASCII characters.

Air OS main screen
Air OS main screen

The system requires an access point, which is configured for the studio side making the transmitter site stub network the station side.  The access point is configured not to advertise its SSID, thus it should be transparent to anyone sniffing around.  The WLAN is configured as a layer two bridge, which will cut down on the data overhead, as layer three framing will not need to be opened between the two units.  The transmitter site network is set up with SOHO router function built into the Nanobridge.  One static route is needed to get to the main network.  Once the security cameras are installed, PAT may need to be used to access individual camera units via the public network.

Ubiquity air os signal strength screen
Ubiquity air os signal strength screen

Next step, deploy the units and aligning antennas.  These are 22 dBi gain antennas, which have a pretty tight beam width.  Maximum transmit power is 23 dBm, or 200 mW.  The transceiver/antenna unit has a handy signal strength meter on the side of the unit, which is good for rough in.  The web interface has a more precise meter.  In addition to that, there is a java based spectrum analyzer, which is very handy for finding open channels in congested areas.  These units can also be used on UNii frequencies with special requirements.

According to the manufacturer, UV-resistant shielded Category 5e cable should be used for outdoor installations.  We have several spools of Belden 1300A, which fits the bill.  The shielded Cat 5 is necessary for lightning protection as the cable shield offers a ground path for the antenna unit.  The antenna mounting structure is also grounded.  I did not take the equipment apart to examine, but I believe the POE injector and antenna have 15KV TVSS diodes across all conductors.  It will be interesting to see how these units do at the transmitter site, where there are two 300-foot towers that likely get struck by lightning often.

More pictures of the installation when it is completed.

Next step, put the system into service and monitor the link.  At the transmitter site, a re-purposed 10/100 Ethernet switch will be installed for the cameras, computer, IP-RS232 converter, and anything else that may need to be added in the future.  One thing we may try is an Audio of IP (AoIP) bridge like a Barix or Tieline for program audio and room audio.

The Relentless Drive to Consolidation

In this blog post about the NAB radio show, Paul McLane (Radio World editor) discusses the reduction of technical people in attendance at the conference.  Consolidation has brought about many changes in the broadcasting industry, engineering has not been immune to these changes.

Because of consolidation, engineering staff has been reduced or completely replaced by contract engineering firms.  Since the Great Recession of 2008-09, this trend has picked up speed.  Expect it to continue to the point where large broadcasting companies employ one engineering staff administrator at the top, and several regional engineering supervisors in the middle, and the bulk of the work performed will be done by regional contract engineering firms.

There is no reason to expect the media consolidation process to stop any time soon.  It will continue in fits and starts depending on the congressional mood and the awareness or lack thereof of the general public.  The NAB itself seems bent on removing all ownership regulations and eventually, with enough money spent lobbying Congress, they will get their way.   Thus, the majority of radio stations will be owned by one company, the majority of TV stations will be owned by another company and the majority of newspapers will be owned by a third.

There will be some exceptions to that scenario; public radio and TV, privately owned religious broadcasters, and single station consolidation holdouts.  If funding for public radio and TV gets cut, which is very likely if the economy collapses further, they will be up for grabs too.

Cloud based network diagram
Cloud-based network diagram

For the future of radio and radio engineering, I see the following trends developing:

  1. National formats will be introduced.  Clear Channel already does this somewhat with its talk radio formats.  Look for more standardization and national music formats for CHR, Country, Rock, Oldies, Nostalgia, etc.  These were previously called “Satellite Radio” formats but I am sure that somebody will dust off and repackage the idea as something else.  They will be somewhat like BBC Radio 1, where a single studio location is used with local markets having the ability to insert local commercials if needed.  Some “local” niche formats will still exist and major markets where the majority of the money is will continue to have localized radio.
  2. Audio distribution will move further into the Audio Over IP realm using private WANs for larger facilities, and public networks with VPN for smaller facilities.  AOIP consoles like the Wheatstone Vorsis and the Telos Axia will become the installation standard.  These consoles are remotely controllable and interface directly with existing IP networks for audio distribution and control.  Satellite terminals will become backup distribution or become two-way IP networked.
  3. Cloud-based automation systems will evolve.  File and data storage will be moved to cloud base servers using a Content Distribution Network topology.  Peers and Nodes will be distributed around the country to facilitate backup and faster file serving.
  4. Continued movement of the technical operations into a corporate hierarchy.  Technical NOC (Network Operations Center) will include all facets of facility monitoring including transmitters, STLs, automation systems, office file servers, and satellite receivers via IP networks.  The NOC operators will dispatch parts and technicians to the sites of equipment failures as needed.
  5. Regional contract engineering and maintenance firms will replace most staff engineers in all but the largest markets.  Existing regional engineering firms will continue to grow or consolidate as demands for services rise.  Those firms will employ one or two RF engineers, several computer/IT engineers, and many low-level technicians.
The most important skill set for broadcast engineers in the coming five to ten year period will be IP networking.  Everything is moving in that direction and those that want to keep up will either learn or be left behind.