ATIS 0500030-2016 Guidelines for Testing Barometric Pressure-Based Z-Axis Solutions.pdf

《ATIS 0500030-2016 Guidelines for Testing Barometric Pressure-Based Z-Axis Solutions.pdf》由会员分享，可在线阅读，更多相关《ATIS 0500030-2016 Guidelines for Testing Barometric Pressure-Based Z-Axis Solutions.pdf（17页珍藏版）》请在麦多课文档分享上搜索。

1、 BEST PRACTICES ATIS-0500030 Guidelines for Testing Barometric Pressure-Based Z-Axis Solutions As a leading technology and solutions development organization, the Alliance for Telecommunications Industry Solutions (ATIS) brings together the top global ICT companies to advance the industrys most pres

2、sing business priorities. ATIS nearly 200 member companies are currently working to address the All-IP transition, 5G, network functions virtualization, big data analytics, cloud services, device solutions, emergency services, M2M, cyber security, network evolution, quality of service, billing suppo

3、rt, operations, and much more. These priorities follow a fast-track development lifecycle from design and innovation through standards, specifications, requirements, business use cases, software toolkits, open source solutions, and interoperability testing. ATIS is accredited by the American Nationa

4、l Standards Institute (ANSI). The organization is the North American Organizational Partner for the 3rd Generation Partnership Project (3GPP), a founding Partner of the oneM2M global initiative, a member of and major U.S. contributor to the International Telecommunication Union (ITU), as well as a m

5、ember of the Inter-American Telecommunication Commission (CITEL). For more information, visit www.atis.org. Notice of Disclaimer i.e., no need to evaluate how the intricacies of the compensation scheme elements perform. Thus, no separate measurement of intermediate results from various error sources

6、 is required. Use of the testing guidelines, outlined in this document will reflect the combined effect of the top three error sources, which are weather, device bias, and indoor building effects, such that results can be extrapolated to areas outside of the testbed, provided a comparable deployment

7、 of the compensation system is present in those areas. Specific attributes of these top three error sources are described below and should be taken into account in test bed planning. 5.1.1 Weather Effects The specific test parameters for randomization during testing are: Full range of barometric pre

8、ssure experienced during a year. o Nominal value is 1013.25 hPa, with values greater than 1020 being high, and values below 1010 considered low. o However, low versus high depends on relative geographical distribution of levels (e.g., a value of 1020 can be low, because it is surrounded by a ring of

9、 1025 values). Full range of wind conditions as a measure of stable or unstable conditions. o Stable conditions with no wind are usually associated with high pressure. Unstable conditions with varying wind speeds are more associated with low pressure. o Strong winds can cause pressure variations ins

10、ide of buildings. Full range of temperature conditions experienced during a year. o San Francisco can be tested year-round due to its limited temperature variation and large wind fluctuations. o Atlanta should be tested in the winter and, if practical, in the summer months as well to assess larger i

11、ndoor/outdoor temperature differential. o A few test sites in a northern city, such as Chicago during extremely cold weather, may be necessary to supplement the testing in Atlanta and San Francisco. Test schedule planning is required for the general season (winter versus summer), but weather changes

12、 over a multi-week collection cycle are expected to vary enough to provide sufficient randomization. The recommended approach is to observe and record the general weather conditions as either stable or unstable, and confirm that there has indeed been sufficient representation of both conditions duri

13、ng testing. Particular attention to results from tall (over 10 story) sealed buildings in both calm and high wind conditions should be tested. Upon completing the test program, if assessing the contemporaneous record reveals a particularly benign weather period, then consideration should be given to

14、 repeat testing scheduled to coincide with expected unstable conditions. Reference data collection from weather stations in the test area should include: Barometric pressure level and pressure tendency (stable, rising, falling), ATIS-0500030 5 Wind speed (average, min, max), and Temperature (min, ma

15、x, average). Weather reference network density, frequency of update, and accuracy of surveyed reference altitude all impact correction accuracy in a compensated barometric z-axis system. For test results from the test bed to be extrapolated to areas outside of the test bed, at a minimum the compensa

16、tion network vendor or wireless carrier must be able to certify that these three compensation factors are consistent with those present in the test bed locations. On a more general point, other compensation complexities, including temperature, humidity, wind, building databases, etc. may also impact

17、 the accuracy of compensation, and no test bed results can be extrapolated unless the vendor or carrier provides assurances that comparable compensation techniques and network capabilities are present in the non-testbed areas. 5.1.2 Handset barometric pressure device bias and drift attributes Bias i

18、n the pressure sensor device contained in the handset will show up as a consistent error in height (high or low) compared to the surveyed truth. Randomization across the following parameters will aid characterization of this effect: Use multiple handsets during a single test, and compare the results

19、 from different models. Use handsets with pressure sensors from at least the dominant sensor vendors (Bosch and STMicro) which account for the majority of market share in the US. Other sensor manufacturers should be included if available. Use handsets across price tier (high and low) as a proxy for

20、the sensors being calibrated at the factory or not. Use handsets available today in volume. Test multiple handsets at the same time, so phone bias in each case can be separated from other error sources. Some proportion of test handsets should include phones manufactured at least one year prior to th

21、e test in order to identify whether aging or drift significantly impact bias and produce outlier results. Phone operating mode may affect barometric pressure values. Some indications in the literature and various experiments show interactions between the barometric pressure measurement and other pho

22、ne sub-systems. It is suggested that the handsets be operated simulating an emergency 911 call. Specifically: Dialing a 911 call simulator to activate GPS and any other location technology systems or heat generating components resident on the handsets should appropriately account for these effects d

23、uring testing. This testing approach is consistent with current x-y testing approaches which simulate a 911 call when taking x-y location fixes. 5.1.3 In-Building Effects Pressure inside of a building can be different than pressure outside it due to HVAC (Heating Ventilation & Air Conditioning) from

24、 fans forcing air into or out of the building. Additional pressure differential due to temperature and moisture differences between inside and outside of a building is known as the stack or chimney effect. The stack effect is most pronounced when the outside air is cold and very dense, while the air

25、 inside of the building is heated and less dense. Cold air flows into the building and the warm air rises within the building. This can cause pressure differences between inside and outside of the building as a function of building height, architectural ventilation zones, and how well sealed the bui

26、lding is. ATIS-0500011, ATIS-0500013, and ATIS-0500022 address building variability from an RF penetration perspective, but not a barometric pressure perspective. In addition to the building variability selection criteria specified for horizontal location testing, acceptable vertical location testin

27、g requires that test bed building selection include an appropriate representation of HVAC managed (sealed) versus un-managed buildings, particularly for taller test buildings 10 stories or greater. This can largely be accomplished by including both newer construction and older construction in the te

28、st buildings selected. ATIS-0500030 6 5.1.4 Test Point Selection, Sample Size, Test Duration, and Instrumentation The generalized instructions for site selection in ATIS-0500011 and ATIS-0500013 should provide the following randomization already: Rural, Suburban, Urban, Dense Urban. Building Height.

29、 o Low RF penetration Loss 1 to 2 story. o Moderate RF penetration loss first or second floor of a 4 to 8 story building. o High RF penetration loss underground parking lots, inner offices of high rise buildings. ATIS-0500013, clause 5 defines: Residential (single family, multi-family 2-4 story, and

30、 multi-family multi-story). Commercial (1-2 story, multi-story). Number of test points range from 1 to 6 depending on building complexity. ATIS-05000013, clause 6 describes the planning process for Number of Test Points and Time interval between Test Calls: 2 test points for a single family residenc

31、e, to 6 test points (low/medium/high x periphery/core) for a tall building. 30 to 100 test calls. Call duration 30 seconds or until complete with a typical test site taking approximately an hour, although due to pauses required between test calls this duration could be two hours. This number of test

32、 points and calls per test point are expected to sufficiently capture both the errors due to weather compensation and handset device bias effects. The sample size of 30 to 100 independent samples (test calls) recommended for x-y technologies is large enough to describe both the mean and standard dev

33、iation of the sensor over the observation period. In contrast to x-y location technologies that operate indoors with outdoor emitters and suffer corrupted signals due to absorption, reflection, and diffraction and often have high positioning error variance, the variance in height measurement of a ba

34、rometric-based device per call at any given point is expected to be quite small relative to the height of a single floor. Each individual barometric pressure report appears to be independent at the handset. There is some low pass filtering in the sensor that removes high frequency noise, but the tim

35、e duration of the filter is short compared to the typical call duration of 10 to 30 seconds. This means that there is no sharing of information between each barometric pressure report across multiple calls, rendering the samples independent. The duration over which the 30 to 100 samples are collecte

36、d ideally should capture the device behavior and the changes within a building, as well as weather dynamics. These longer duration effects can effectively be captured by using the whole body of collected data amongst various buildings and floors, in a variety of quiet and unsettled conditions. To co

37、nfirm this basic test plan tenet, one or two test sites may be run for 12 to 24 hours, with all test handsets running simultaneously, to confirm there is no phone-specific variances or combinations of weather or indoor conditions that would only be seen in a long duration data collection. Data resul

38、ts from a 16-hour overnight collection in unsettled weather conditions supports the idea that all the handsets show consistent behavior over long periods with no change in mean or standard deviation of z-axis error as a function of observation interval. In-building HVAC and stack effects, however, m

39、ay cause pressure differences that vary up the vertical building profile in an unpredictable manner. Thus, some additional test points should be used to assess whether this variation is sufficiently captured by the “low/middle/high” floor test points, or if larger variation is present and additional

40、 testing is required in certain building types (primarily tall sealed buildings). This additional testing can be fewer (e.g., 20) fixes, every couple of floors (every other floor or every 3rdfloor), possibly using interpolated ATIS-0500030 7 vertical ground truth to minimize professional survey cost

41、s. The average phone height error plotted against floor number should reveal any floor level correlated HVAC or stack effects that would not be revealed by just the (low/medium/high) test points. If the low/middle/high variations reasonably capture the variations revealed in the more granular test s

42、amples, then ongoing vertical testing may rely solely upon the low/medium/high test points utilized for horizontal testing. It is advantageous for subsequent data analysis to have an accurate measurement of the ambient barometric pressure at the test point. It is therefore recommended that the test

43、setup (the test handset tray or cart) be equipped with a lab quality calibrated barometer. One example that was used effectively in recent Qualcomm tests is Model 745 from Paroscientific Inc. The readings from this reference barometer should be recorded at each test point during the testing. 5.1.5 R

44、ural Testing Consideration and 80% of population across top 50 CMAs Carriers will need to determine if vertical height reporting is required in rural areas of the top 50 CMAs to meet the 80% population coverage requirement. To support this determination, the vertical yield of all systems under test

45、should be collected for all morphologies and all test points. Rural environments are expected to have only 1 to 2 story buildings, and thus height reporting is less important in emergency situations. There are some unusual outdoor search and rescue scenarios where altitude reporting would aid in a G

46、PS/RF shadowed valley, however those fall outside of the FCCs 4thReport and Order mandate and are not addressed here. ATIS-0500030 8 Annex A (informative) 6 Annex A Background Material on Various Error Sources The following clause provides background material on weather and various barometric pressu

47、re error sources. The discussion is meant to be informative rather than prescriptive. Temperature differences cause regions of low and high pressure A primary weather effect is caused by high versus low atmospheric pressure. As air warms, it rises, causing a low pressure region at the surface. When

48、air cools, it descends causing a high pressure region on the surface. Figure 1 High Low Pressure Areas5Wind6flows from areas of high pressure to areas of low pressure7. This is due to density8differences between the two air masses9. Since stronger high-pressure systems contain cooler or drier air, t

49、he air mass is denser and flows towards areas that are warm or moist, which are in the vicinity of low pressure areas in advance of their associated cold fronts10. The stronger the pressure difference or pressure gradient, between a high-pressure system and a low-pressure system, the stronger the wind.115This image is available from meteoblue, Basel, Switzerland at: 6Available at: . 7Available at: . 8Available at: . 9Available at: . 10Available at: . 11Available at: . ATIS-0500030 9 Annex B (informative) 7 Annex B: Tabulation of Err