Cellular Infrastructure and Standards while deploying an RDA Overview This whitepaper discusses the methods used while deploying an RDA into a field environment and dives into the standards used to judge the apparent success of the RDA in acquiring and distributing a cellular signal. This paper will only discuss the RDA units which contain either AT&T or Verizon Wireless cellular antenna. LTE is the industry preferred high speed data technology for the foreseeable future and is being deployed by all of the major cellular carriers in the U.S. and around the world. For that reason this paper will only focus on LTE signal despite the RDAs ability to acquire usable 3G signals as well. Each vendor uses their own spectrum band for LTE which means each is on their own frequency to avoid interference. These bands are important for Motion to understand so that the antennas could be designed around the center frequency of each vendor. AT&T operates on band 4 and 17 which means they are centered on 1700 and 700 MHz Verizon operates on band 13 which means they are centered on 750 MHz Using this knowledge Motion has been able to create antennas matched perfectly around each vendor frequency. The primary feature of the cellular RDA deployed in the field, whether this is a large outdoor construction site or a small indoor warehouse, is to provide internet connectivity to the end users. They are then able to use the connectivity provided for everything from communicating with each other via Skype or keeping track of payroll and accounting. How to measure success A problem that is often seen in the wireless industry is how to judge the overall success of deployed wireless devices. It is the same problem that Motion and its customers have; how do you judge success or even improvement in the quality of signal, bandwidth and throughput. In order to determine the success rate at which the RDA is providing this always on connectivity, Motion has developed a way to quickly analyze the performance of the radio signal and therefore overall performance at the time of deployment. This methodology has allowed Motion to successfully locate the ideal placement of the RDA unit in the field. This allows for a quick set up; enabling the users to access the RDA provided network in a short amount of time all while achieving the maximum bandwidth and therefore user throughputs in that location. The first step in understanding these methods is knowing the parts of the wireless infrastructure involved during an RDA set up. This paper breaks the infrastructure down into two parts, the OEM cellular side and the Motion WAN side. Page 1
Next, an understanding of the tools used to analyze wireless networks is necessary. This paper will discuss various testing methodologies that are used by the end user such as the industry standard speed test and why it must be combined with proper network diagnostics to fully understand the wireless network s health. Successfully applying this knowledge has enabled Motion to achieve total success in setting up wireless networks at customer sites across the United States and Canada Standard Wireless Infrastructure Components Wireless infrastructure can be broken down into two components and understanding the limitations of each of these components is the key to setting up an efficient, high speed wireless network. Chart 1 below is a block diagram of the infrastructure breakdown used in this paper. Two parts of wireless infrastructure Cellular Side WAN Side Laptop 1700/700 MHz RDA 2.4/5 GHz Smart phone BTS (3G) or enodeb (LTE) RDA Tablet Chart 1: Wireless Infrastructure block diagram The first component is the cellular side. This consists of the cellular provider s network which includes not only the physical 3G base stations (BTS) and 4G LTE enodebs but the frequency and bandwidth provided by the cellular company which is received by the RDA. The enodeb is the equivalent of a base station for a 3G antenna and is the hardware that connects the received LTE signal from the antenna to the backhaul network and therefore the internet. The cellular side operates at either 1700 or 700 MHz depending on the provider as discussed in the section above. Page 2
This part of the infrastructure cannot be changed or modified during set up of the RDA and is therefore out of the hands of the engineer who is installing the unit. This is important to understand because the RDA will not function in the same capacity at every location but rather will only be able to take advantage of what coverage the service provider has allowed. For example, faster Wi-Fi throughputs will be observed in an open area with direct line of sight (LOS) to the cell tower compared to a downtown municipal area that is surrounded by large metal buildings. In order to better understand the different environments the RDA could be placed in, Motion has tested it in many locations such as direct LOS in Fresno, California, downtown areas in Denver, Colorado, and indoor locations in Baltimore, Maryland. These tests have enabled Motion to establish baseline numbers which will be discussed in a later section. Using the results from these tests, Motion has developed a way to improve performance in highly congested areas as well as low coverage areas with poor RF conditions. This would include areas such as downtown city centers, (with very high noise in the LTE frequencies), by using solutions such as 10 db highly directional Yagi antennas instead of the usual omnidirectional dipole antennas. These however are also still limited by the RF conditions and environment of the installation zone. The graphic below illustrates a situation where low throughputs may occur simply due to the placement of buildings and the cellular towers and therefore the signal power reaching the RDA will be extremely low. Figure 1: Low RF conditions example Page 3
The second component in a wireless infrastructure is the WAN side. This includes the RDA unit and the 2.4 or 5 GHz Wi-Fi network it deploys that is then received by the user s device whether that is a mobile phone, laptop, tablet or second RDA unit. Since this is the component of the overall wireless infrastructure that can be modified, it is the key to correctly deploying the RDA and when fully understood allows the RDA installer to achieve maximum speeds or throughputs in the area. Although not always possible, the ideal position for the RDA is in LOS of an enodeb enabled cellular tower. Since many times this is not possible, as referenced earlier Motion has developed a highly directional antenna system which can be used to boost the RDAs range. This Yagi antenna system replaces the onboard cellular antennas which eliminates the omnidirectional behavior of the RDA and instead creates a pencil beam radiation pattern that has extremely high gain. This system is especially useful when trying to reach a tower outside of the standard antenna s range. When a nearby cell is too congested, this method is an excellent way to boost the user throughput speeds by three or four times the usual speed. Figure 2 below shows an RDA unit attached to the Yagi antenna system. The ray dome is removed simply for testing purposes. It also shows the difference in the gain pattern when the omnidirectional antennas are replaced by the directional Yagi. Page 4
Figure 2: RDA unit with Yagi antenna and omnidirectional vs. Yagi gain pattern Wireless network performance tests Speed test The first quantitative statistic used when testing a wireless network is the network speed or throughput. This is an important test as it is a fairly good judge of what the end user s network experience will be however if a network is judged solely based on this test, the network s overall health may be misdiagnosed. The most popular site used to test throughput is known as the Ookla speed test and can be found at www.speedtest.net. This test determines the user s IP address and then locates the closest server to that relative location. It then sends test packets to and from the user device to determine ping time, download, and upload speeds. Page 5
Figure 3: Screenshot of Ookla Speed test and results This test is a fairly good indicator of what speeds the user will experience when downloading basic files at that time of day however it has many flaws. The first flaw associated with judging a network s health by the speed test is the inability of the user to know if these are the true maximum speeds being seen by the RDA or if they are capped by the vendor. This capping process is very common with cellular service providers and refers to the hard limit they set on bandwidth provided to each device connected to the network. This process is especially common in areas with high cell congestion. The purpose of this bandwidth capping is to create an equal experience for every user rather than providing certain users large amounts of bandwidth and others almost none during periods of high cell usage Although this is a good practice in theory, it makes judging a network s health very difficult if only relying on this test. The figure below illustrates an example of this capping process. This was a speed test recording taken during testing on an RDA unit in California. Despite the download speeds averaging greater than 10 Mbps, the upload speeds are stuck around 2 Mbps. This is a case of provider capping since the unit will see equivalent uplink and downlink speeds in normal situations. Figure 4: RDA Speed test example done in California Another inaccuracy of the speed test occurs due to the location and speed of the servers used by the speed test company. Looking at Figure 4, it can be seen that the server being used to test the speeds is located around 100 miles away. This is purely an estimate based on the user s IP address and is Page 6
therefore not very accurate which is an issue as generally a user located physically closer to the test server will have lower ping times and therefore superficially appear to have a better connection. The speed or power of the pinged server can also cause a misjudgment of the network s health. This can be observed on a fairly consistent basis by running a ping command out of the Windows command prompt on both Google and Yahoo servers. The Google response time tends to be around 10ms faster due to the difference in their server processing power and not due to the wireless network. A final inaccuracy of this speed test is due to the time of day at which the test is run and the overall congestion of the cell the device under test is attached to. An example of this is if the cell is in the area of large office buildings. The test will show low throughputs during the business day and grow increasingly faster as the business day ends because the number of people using that cell will diminish. Wireless performance tests RDA diagnostics This section will discuss a more accurate tool to judge the overall health of the wireless network being used and output by the RDA. This tool can be found by logging into the RDA unit s admin panel while connected to the RDA produced network. The default IP address used is 192.168.1.1. A screenshot of the diagnostic panel is below entitled Figure 5 and can be found under the custom commands section. Page 7
Figure 5: Diagnostics tool screenshot Using this tool, the user can verify that they are connected to the LTE by checking for the notation Successfully got signal info LTE: after running the highlighted command WANdiagnostic. The custom command screen can be found by setting the mouse over the highlighted system tab. Next run the command WAN-diagnostic. These four statistics, RSSI, RSRQ, RSRP, and SNR, can be referred to as key performance indicators (KPIs) and will be discussed in detail in the next few paragraphs. The first KPI to look at when determining RDA wireless network health is RSSI this stands for receive signal strength indicator. This measures the power of all the LTE reference subcarriers and an example can be seen below in Figure 6. This is a screenshot of a spectrum analyzer showing power versus frequency and is only showing the LTE reference subcarriers. It is shown using a Sprint LTE signal since they only use 100 reference subcarriers which makes the chart more readable then the 1000 Verizon uses. All the yellow area is the power being measured and included in the RSSI. Figure 6: RSSI measurement: all subcarriers measured The second KPI is RSRP and similar to RSRQ is specific to LTE. This stands for reference symbol received power. Although similar to RSSI, this measures the average of the subcarriers rather than the overall power. Figure 7 shows an example of this which is again a screenshot taken from a spectrum analyzer. Verizon typically deploys 10 MHz channels with 1000 subcarriers which means on average the RSRP measures 30 db lower than RSSI. Page 8
Figure 7: RSRP measurement: subcarriers averaged, effective area in red The next KPI is RSRQ which stands for reference signal received quality. The formula for this KPI is (N x RSRP)/RSSI where N is equal to the number of reference subcarriers (in the case of Verizon; N=1000). The final performance indicator that is used in this diagnostics screen is SNR which stands for signal to noise ratio and is simply the logarithmic value of the signal power divided by the overall noise power. An important thing to note here is if the SNR is too low, the diagnostic tool will not give an accurate number which may result in a negative SNR. These KPIs allow a much more thorough examination of the network s health than the typical Speed test application. An example of some of these readings which were taken using an RDA in California are below and give a good indication of the network health. In these tests the RDA was tested using different antennas one of which included highly directional Yagi antennas. For this reason we expected much different readings however during all these readings the speed test application was producing around the same numbers (see Figure 3). KPI Broadband Antenna 700 LTE only Directional Yagi RSSI (dbm) -38-35 -11 RSRP (dbm) -65-64 -44 RSRQ (db) -10-10 -10 SNR (db) 22.2 17.4 7.4 Table 1: California RDA readings Clearly the directional Yagi antenna was the most effective antenna setup (in this case the lower number is better since it is using logarithmic scale) but if the test had only been done using the speed test; no true decision could have been made due to the carrier bandwidth capping. The diagnostics also enable Motion or the RDA installer to view what kind of RF conditions the unit is being placed in. Table 2 shows real life readings taken in different areas. Page 9
KPI Dallas (indoors) Fresno (outdoors) Denver (downtown) RSSI (dbm) -63-59 -51 RSRP (dbm) -92-82 -83 RSRQ (db) -11-9 -16 SNR (db) 3.2 9.8-3.4 Table 2: RF Conditions in different environments This table is a good reference to consult when setting up an RDA. The downtown area produced about the same signal strength as the outdoor environment taken in Fresno, California but the noise floor was so high due to the many reflections and signal dispersion the RDA diagnostics were in error and caused a negative SNR value. This highlights the importance of looking at all of the KPIs and not just one or two. The chart in Figure 8 may be a good reference to use during set up to determine which type of RF coverage is available in the area and takes advantage of the discussed diagnostic performance indicators. Figure 8: RF condition reference chart Results Summary The problem confronting Motion and its partners was to find the best position and orientation of the RDA in a short amount of time to enable quick installation and setup while achieving maximum user throughputs. This is important because the performance of the RDA is maximized when placed in the best RF conditions in the area. Most people want to use the standard speed test application to find this location. This provides a fairly good indicator of user experience however does not help in determining the best long term position and location as it may give false readings due to cellular service provider capping and time of day or cell congestion issues. The solution to this issue is to use other metrics that can be found in the RDA onboard diagnostics mainly the four key statistics called KPIs (RSSI, RSRP, RSRQ, and SNR). Using these four KPIs it is possible to tell the best setup location and orientation for the RDA in a short amount of time. Page 10
The Motion Wireless and LINCWorks team is available for support and more than willing to schedule a conference call or onsite visit to help in the process outlined in this paper. We are fully committed to our customers and the success of the LINCWorks products. Please contact myself or Dave Mackie at the below addresses. Paul Peterson Wireless Engineer ppeterson@motioncomputing.com Dave Mackie Head of Wireless Network Products dmackie@motioncomputing.com Page 11