The following guide will help you set up a Ubiquiti Bullet as a WiFi base station, including best practices for surge protection, and cabling.
The Ubiquiti Bullet series are a high powered 2.4GHz or 5.8GHz outdoor radio (denoted M2 or M5 respectively). With a cylindrical "bullet" shaped form factor the device is designed to sit inline between an Ethernet cable input and a WiFi antenna output. Being completely independent of antenna choice provides a high degree of flexibility during integration, allowing you to not only swap antenna to best match surroundings, but also to hide the radio from view, or protect it from harsh/hazardous environments.
Like all Ubiquiti equipment the Bullet is capable of being set in any wireless mode - access point, station, bridge, router, AirMax (for P2P/P2MP) AP/station. In this example we're setting the Bullet up as a high powered 600mW WLAN to allow any laptop, mobile phone, etc to access the local network & Internet. While this configuration is quite universal, in the demonstration we're setting up this gear on a tripod - similar to what we've provided for many government & research teams setting up temporary wireless access in remote sites. The following list of equipment was used in this example:
While the Bullet has full router capabilities the majority of us will not be using it in this manner, so to demonstrate it's capabilities we're setting the unit up as an Access Point in Bridge mode, so packets will be passed to the Dovado router untouched (no NAT, DHCP, etc).
Naturally choosing an antenna depends on the desired application, the most common consideration being the desired coverage area. Choosing a directional antenna over a non-directional (omnidirectional) antenna should be self explanatory - to provide a stronger signal in one direction by trading off for a weaker signal in other areas. However antenna gain isn't so obvious, and also an important consideration given Australia's strict maximum power limits (EIRP). If you've read some of our other antenna guides you'll know that an increase in gain (dBi) can only be achieved by focusing signal - in the case of an omnidirectional antenna signal is focused outwards horizontally, reducing signal power transmitted above and below the antenna.
There's a common misconception that the 2.4GHz ISM band is "unlicensed" in Australia, where in reality the ACMA regulates it with a 'Class Licence" which basically states that anyone can operate on the band provided they stick within the conditions imposed by the ACMA such as obeying power limits. The maximum EIRP limit for 2.4GHz in Australia is 4000mW (36dBm). The real world implication of this is the Bullet M2 has a 600mW (~28dBm) output power so the maximum antenna size we can use is an 8dBi antenna (well, depends on cable losses).
Like all of Ubiquiti's PoE powered equipment the Bullet has only a single Ethernet input. The waterproofing gland is slid over the cable, and then cable inserted in the base of the Bullet. Very simple.
The other end of the Ethernet cable was then connected into the PoE injector, which was in turn connected into the Dovado Pro (with 4G USB modem) to provide WAN + LAN connectivity. WLAN was switched off from the Dovado to limit channel interference, due to the presence of several other WiFi access points in our test location. If we did not have a suitable router we could have instead connected the Bullet directly to the laptop and allowed the computer to share the 4G connection to the Ethernet port. In this instance we would have set the Bullet in 'Router' or 'SOHO Router' mode instead of Bridge mode.
The 6dBi RFI COL2406 antenna was chosen as it tends to be a good all-rounder - durable, low profile, good omnidirectional pattern, and stud mounted making it easy to attach on any 13-16mm hole diameter bracket.
While the base of the Bullet is sealed with its supplied cable gland, it's important not to overlook the N-Type connector on the top of the radio which requires sealing against moisture ingress. Self-amalgamating tape such as rubber-butyl "Coax-Seal" is an ideal choice of sealant tape. When applying it's important to keep the tape under tension, and overlap each wrap by about 50% of it's width to ensure a good seal. It's important to begin sealing the connection from the cable all the way to the top of the radio - any part of the N connector from the crimp ferrule to the radio exposed has the potential to be a point of ingress. After completely wrapping the connection work the tape with your fingers to eliminate any gaps between the overlapping folds.
The effects of improperly weatherproofed connections can be minor to severe, the most common symptom being a change in impedance resulting in higher VSWR, which results in less power being transferred from the radio to the antenna. In the lower layers this can reveal itself as a reduction in modulation rate or a reduction in code rate as link adaptation compensates for lower channel power.
Don't risk your radio! Network continuity should not have to be disrupted every time there's a chance of storm activity.
While no form of surge protection will protect against a direct strike, a good quality coaxial arrester can protect against the devastating effects of an electrical surge. The majority of damage to electrical devices are the result of the electromagnetic pulse (EMP) generated from a nearby lightning strike, which parasitically couples onto any vertical/near vertical metal object (cables in particular), often called 'lightning transients'.
Coaxial surge suppressors/arresters (the terms are interchangeable) should be installed as close to the radio as possible. In the case of our installation where the Bullet is mast mounted we have only a short tail of coax to worry about, so only a single arrester is required. If however we were hiding the radio in a building or equipment cabinet and had several metres of coaxial cable then a secondary arrester often called the "tower-top arrester" should be installed between the antenna and feeder cable. Lengths exceeding 30m may also benefit from a tertiary arrester or mid-span grounding kit being installed half-way down the feeder.
The below images show a typical installation when using a LMR400 feeder cable (ignore the hose-clamps, I've just set it up temporarily for the demonstration).
Now that the hardware has been completely installed, it's time to configure the Bullet to act as a WiFi Access Point.
Configuration is performed using your web browser by typing in https://192.168.1.20. You'll get a warning pop up stating that the device has presented an invalid SSL certificate due to the certificate being self-signed and not issued by a CA. This is normal so just hit accept/continue. You'll then need to log in with the default username and password of "ubnt".
The first step is to configure the Wireless tab so the Bullet acts as a WiFi Access Point. Change the settings according to the below picture. Be sure to set the device up with WPA2 security and an appropriate WiFi password here. You can also use this tab to adjust the power output should you wish to cover a smaller area. The default channel bandwidth is 40MHz, but some older devices such as wireless printers and mobile phones cannot operate on 40MHz, so we've changed it to 20MHz in this setup. WDS mode is not required in this instance.
Navigate to the left-most tab (with the Ubiquiti logo) and uncheck the airMAX checkbox to disable Ubiquiti's proprietary airMAX protocol (a special MIMO TDMA multiplexing technique that allows communication between two or more Ubiquiti devices). You won't be able to untick the box until the Wireless tab has been completely configured. AirMax has the ability to increase wireless channel usability from 30% (on 802.11n) to as high as 90%.
In this case our Dovado router is located at 192.168.0.1 with a subnet mask of 255.255.255.0, meaning all our devices are restricted to the 192.168.0.XXX subnet. This means that we've had to change the default 192.168.1.20 to 192.168.0.20. As we want the Dovado to continue controlling the network we're setting the Bullet to Bridge mode, and then assigning the Default Gateway and DNS IP addresses to the address of the router. We've also switched off the Dovado's WLAN just to free up some spectrum as we've already got a collocated ADSL WLAN which wouldn't have left much space for the Bullet.
To assess performance we're going to test the Bullet M2 versus the more substantial Rocket M2 setup. Our location in a mixed commercial/residential area is an ideal proving ground to demonstrate resistance against co-channel interference (ie collocated AP's, residential WiFi, etc), multipath propagation through dense urban environment consisting of both trees and buildings.
Both the Bullet and Rocket were mounted (at separate times) on the same mast which was positioned indoors at the roller-door entrance of our warehouse. A simple walk-test was conducted using a simple program called inSSIDer which provided dBm readings of both our test AP and neighbouring (interfering) AP's, along with good old fashioned command prompt to provide some basic packet loss statistics. Performing a ping test is a great simple way to test link conditions - ICMP does not have a guaranteed delivery mechanism so deteriorating radio conditions are instantly obvious. Ideally a more comprehensive WiFi test should be conducted with JPerf/IPerf but here I'd prefer to demonstrate some basic stats that can be collected on your home network without requiring any extra tools.
The Bullet M2 had a stable range of about 150m through a typical suburban non-LOS environment. Range extended out to about 200m if higher packet loss was tolerable (11% loss @ 194m) - this sort of loss might be acceptable for a simple telemetry station, but not for bandwidth intensive applications. With a range about 400% greater than a normal WiFi access point this is quite a good result, especially in the presence of 24 detectable and 6 overlapping access points.
Before we begin it's important to note that this may be a somewhat unfair comparison - the smallest available Rocket M2 base station antenna is 10dBi. To compensate for the higher gain we reduced the dBm output by 4dB, however as the radiation pattern differs this should not be overlooked.
Walking out to the 194m mark where we experienced an 11% packet loss previously, this time not a single packet was lost. It took as walking out to 325m before similar levels were experienced, with a usable range of about 275m. Both maximum and average latency were also significantly reduced - by almost half that of the Bullet.
While this simple demonstration should in no way be considered (and was never intended to be) empirical in nature, it's quite clear that the Ubiquiti Rocket outperformed the Bullet in both range and reliability. While Rocket does not employ smart antenna techniques such as beamforming, multiple antenna technology clearly provided an advantage over a single antenna WiFi radio, not surprising in an environment relying entirely on radio scatter. Ubiquiti Bullet is an ideal compact form factor WiFi solution, suitable for a much more diverse range of applications and environments. However for maximum range and performance, the choice is still Ubiquiti Rocket.