A series of tests will be conducted to determine whether GMS interferes with FMS operations during a FRC event. The venues will be MAR 2015 off-season events. The reviewed results will be published before Dec 14, 2015. There will be two parallel objectives:

Testing to see if GMS can use 2.4 GHz at a venue where the spectrum is crowded. Though GMS will work in such hostile environments, its performance may degrade to a point where users get frustrated.
Testing to see if GMS interferes with FMS in any way in the 5 GHz range.

Document Changes

Version	Date	Initials	Description
1.00	Jul 4, 2015	RP	Initial Submission
1.01	Jul 8, 2015	FWL	Updated formatting Added testing strategies for the 2.4 GHz Band Added request for others to join the effort.
1.02	Jul 11, 2015	RP	Added testing strategies for the 5 GHz Band
1.03	Jul 31, 2015	RP	Updated with comments from Kevin Dieterle

Background

There are 2 radio bands commonly available for WiFi. Each band is partitioned into several channels:

2.4 GHz, Channels 1 to 13 (in USA)
5 GHz, Channels 36 to 165 (though not all channels are permitted in the USA)

The 2.4 GHz band

The 2.4 GHz band is public for Industrial, Scientific & Medical (ISM) use. WiFi shares it with a several other radio technologies, like Bluetooth, ZigBee, WPAN, Wireless USB, etc. In addition, microwave ovens, cordless phones and car alarm remotes use this frequency. Though there is no license required to transmit in this band, all transmitting devices must be certified by the FCC to verify that they meet a set of requirements, like limiting the transmission to the specified channel.

Though there are 13 channels listed, only 3 channels are recommended. They are channels 1, 6 and 11. These three channels do not overlap with each other. Using any other channel is considered 'anti-social' in that it will interfere with multiple other channels, as shown in the figure below. Each channel spans about 22 MHz ("bandwidth") and typically carries about 3 MBytes/sec ("throughput") using WiFi g.

The WiFi protocol is designed to allow peaceful co-existance between multiple WiFi access points within a channel. It is similar to people standing up and speaking into an open mike at a meeting. Each transmitter checks to see if any other transmitter is active before it transmits data. If another transmitter is active, the transmitter waits a bit and tries again. Such politeness usually avoids having transmitters stepping over each others' messages. There is no queue and a transmitter may have to wait and retry multiple times. If there are a lot of transmitters, then a transmitter may have to wait longer to get its turn. Also, if transmitters are sending a lot of data, then they are taking a lot of time at the "open mike" and other transmitters have to wait longer for their turn.

The FCC verifies that a WiFi device configured to operate in a channel will not transmit radio waves in any other frequency. This avoids unwanted interference with other devices operating at other channels.

The Bluetooth protocol allows Bluetooth devices share the band efficiently. Unfortunately, the two protocols are not designed to optimally co-exist with each other since the Bluetooth transmission retry mechanism is different. Also, unlike WiFi, a Bluetooth transmitter is not constrained to one channel in the band, it jumps all over the entire 2.4 GHz band.

The 5 GHz band

The 5 GHz band is also a public ISM range, however, it is a lot less crowded and has much higher capacity. There is no pesky Bluetooth. There are at least 23 non-overlapping channels (compared to only 3 in the 2.4 GHz band.) Each channel spans 20 MHz, but two adjacent channels can be "bonded" together to further increase the data rate. The data transmission technology is newer and faster. The typical data transfer rate for a single channel is about 30 MBytes/sec.

As with 2.4 GHz, the FCC verifies that certified Access Points do not generate spurious radio transmissions. In other words, they certify that an AP configured to operate in a channel will not affect another AP operating in any other channel.

One disadvantage of 5 GHz for WiFi is its relative inability to penetrate solid walls. This limits its range at larger events. The simple solution is to have multiple APs distributed at various points.

Current Status

FMS typically uses one of channels 36, 40 or 44 in the 5 GHz band. Each channel is 20 MHz wide. The AP used is a Cisco AIR-AP1252AG-A-K9, which costs about $250 (refurbished, including the required antennae and fancy power supply.) MAR possesses two of these routers on loan from FIRST.

The main source of interference in the 5 GHz band at events seems to be from "rogue" APs hosted on cell phones. Many iPhones and Android devices allow their owners to create a WiFi "Hot Spot". This allows nearby devices to access the Internet via the phone's data plan. Though it is against FRC rules, there is no practical way of preventing this usage at an event. Assuming that the rouge APs are not hostile, the effect of this interference is small. It is usually some kid trying to show off. Not much data will be transmitted and there are alternate 5 GHz channels that can be used. Willfully interfering with a WiFi signal is not illegal unless the WiFi is for Emergency Responder use.

Most event sites already have a WiFi infrastructure consisting of 5 GHz Access Points. While schools can be convinced to turn off their APs during an event weekend, larger venues (like the Javitz Center in NYC) may be more reluctant. They typically host other events concurrently with an FRC Regional.

Testing Methodology

There will be two separate testing efforts. MAR is fortunate to have a number of different engineering disciplines available to help in this effort. If anyone has time to help please let us know. We are also asking for leads to WiFi vendors that help guide us in this effort.

The 2.4 GHz band

Testing Commercial Gear

It has been suggested that by using “commercial” network radios and routers we may be able to make the 2.4 GHz band usable.

We are looking for help identifying the commercial gear to be tested. If anyone has contacts with other WiFi and Networking vendors we should be using please let us know so we can engage them.

Testing within a saturated venue

Additionally we have asked Stabler to allow us to attend an event at the facility that somewhat mimics our event. We are looking for a similar demographic and believe a concert will fit the bill. There we will put up various WiFi gear to test its saturation limit. Again we are looking for expertise to help in this effort.

The 5 GHz band

A range of components will be tested. While the goal is to use inexpensive Access Points (like D-Link DAP 1522), we will also test with higher end commercial equipment (like the Cisco AIR-AP1252AG-A-K9 used by FMS). Testing will include:

Channel Saturation Testing

This series of tests do not involve either GMS or FMS. Specific software will be written to load the channels and measure various parameters. The tests will be at a venue other than at an event. This allows these extreme tests to be conducted over a longer period of time.

A number of transmitters and receivers will be configured to send large streams of UDP packets within one non-bonded channel. The data transfer rates are expected to greatly exceed those of GMS & FMS. We will plot transfer rates with various statistics, including transit time (max, min, average), packet loss rate, retransmissions, etc. This will give us an idea of the throughput capacity on one channel. We will also use advanced tools to detect and view any extraneous channel transmissions (leakages) by the devices under test.
The above test will be repeated on two non-overlapping channels. The goal will be to measure any cross effect between channels at extreme throughputs. Saturating one channel should have no effect on the transmission capacity of a different channel. If any effect is observed, then the test will be repeated with adjoining and widely separated channels.
The tests will be repeated using inexpensive 5 GHz routers (D-Link) and available commercial routers (Cisco.)
Lastly, we may test with bonded 40 MHz wide channels, though FIRST has considered this option and has decided against it for FMS.

Testing with FMS

GMS can be configured to test sending large files repeatedly between the server and the mobile devices. These tests will involve FMS and GMS and will be conducted at our off-season events.

GMS on a distant channel, with FMS at the opposite end of the band. FMS should not be affected. This test was conducted at the MidKnight Mayhem event on June 27, 2015.
GMS on a channel that is adjacent to FMS. GMS would be on a non-overlapping channel, right next to the FMS channel. FMS should not be affected.
GMS and FMS on the same channel. Clearly a troubling idea. The goal would be to get an idea of FMS symptoms on busy channels. This test will not be conducted during real matches. We will record GMS throughputs and FMS statistics, like packet Trip Time.

If anyone has contacts with other WiFi and Networking vendors we should be using please let us know so we can engage them.

MAR 2015 Off-season Events:

The MAR off season events available for testing are:

May 16 Monty Madness
Jun 27 MidKnight Mayhem
Oct 10 Girl Power
Oct 17 Duel on the Delaware
Nov 7 Brunswick Eruption
Nov 14 Ramp Riot

Results to date

MidKnight Madness 2015

One DAP-1522 Access Point was set up (SSID "staff1") in the event arena about 100 feet from the FMS station. It was configured to use Channel 157 (5 GHz). FMS was concurrently using Channel 36. The school AP ("WWP") was turned off at about 6pm on Friday for the duration of the event. There were no other APs in the 5 GHz range most of the time. Someone had a rogue AP that seemed to come on intermittently on channel 149. The MAC address indicated that it was an iPhone.

GMS was used for Queuing and Match Announcing during the Qualification period. It worked fine. There were no WiFi issues noticed with FMS. Trip times for robot traffic on the field consistently remained below 10ms and no issues with dropped packets were observed.