1、ATIS-0500035 ATIS Standard on Guidelines for Testing Dispatchable Location Alliance for Telecommunications Industry Solutions Approved July 12, 2017 Abstract This document provides guidelines specific to testing and evaluation of dispatchable location within the framework of the 9-1-1 Location Techn
2、ologies Test Bed. It should be viewed as an extension to ATIS-0500031.v002, Test Bed and Monitoring Regions Definition and Methodology. ATIS-0500035 ii Foreword The Alliance for Telecommunications Industry Solutions (ATIS) serves the public through improved understanding between carriers, customers,
3、 and manufacturers. The Emergency Services Interconnection Forum (ESIF) provides a forum to facilitate the identification and resolution of technical and/or operational issues related to the interconnection of wireline, wireless, cable, satellites, Internet, and emergency services networks. The mand
4、atory requirements are designated by the word shall and recommendations by the word should. Where both a mandatory requirement and a recommendation are specified for the same criterion, the recommendation represents a goal currently identifiable as having distinct compatibility or performance advant
5、ages. The word may denotes a optional capability that could augment the standard. The standard is fully functional without the incorporation of this optional capability. Suggestions for improvement of this document are welcome. They should be sent to the Alliance for Telecommunications Industry Solu
6、tions, ESIF, 1200 G Street NW, Suite 500, Washington, DC 20005. At the time of consensus on this document, ESIF, which was responsible for its development, had the following leadership: S. Sherwood, ESIF Chair (Verizon Wireless) R. Hixson, ESIF 1st Vice-Chair (NENA) R. Marshall, ESIF 2nd Vice-Chair
7、(Comtech) J. Green, ESIF ESM Co-Chair (Sprint) K. Springer, ESIF ESM Co-Chair (AT hence, the level of provisioning in the Test Bed regions is assumed to be similar, except for normal regional variations, to that in other CMAs intended to be provisioned in the same timeframe. 4. Testing the quality o
8、f DL entails determining, among other things, its level, e.g., whether a DL Level 1 or DL level 2 or a civic location is delivered and its veracity relative to established truth. Hence the test methodology is to be designed to provide this level of granularity. 5. While the purpose of DL testing is
9、not to troubleshoot, or analyze in detail the performance of each element of the DL ecosystem, a DL test campaign in the Test Bed is an invaluable opportunity to gather as much data as possible to enable refinement of the logical processes or algorithms in the elements critical to the determination
10、of DL. ATIS-0500035 7 6. The results of DL testing will have to be interpreted in the context of the degree to which the NEAD has been provisioned with reference point records and the typical level of detail achievable in those records. It may not be possible, for example, in a DL test campaign to e
11、xamine the effects of details not yet available in the provisioned reference point records, such as Place Type (i.e., classification of building type). It should be noted that the data included in the NEAD database from a given provider belong to that provider, and other parties, e.g., affiliated wi
12、th NEAD management, would not be in a position to add information into such records. Furthermore, the Test Beds central premise is representative testing. Attempting to manually add details in certain NEAD records within the Test Bed that do not reflect processes followed broadly elsewhere would dev
13、iate from that principle. That approach is therefore not followed in the current test methodology. 7. Multiple sequential DL test campaigns may be required to establish the long-term performance of DL as the NEAD provisioning matures, more details are included in some of its records, and the algorit
14、hms of location servers are refined. A DL testing event is not intended to be a pass/fail type of test but rather it is intended to establish DL performance in the given time frame of testing. It is also intended to provide quantitative data and performance insights to wireless carriers and develope
15、rs to support future efforts to improve achievable performance. 7 Dispatchable Location Test Methodology Dispatchable location is inherently different from geodetic (latitude/longitude or x/y) location. DL refers to a certain civic address of a building or a specific unit or specific part and floor
16、within a given building with a civic address and subaddress elements. DL is also affected by different system provisioning and environmental factors from geodetic location. For example, multipath encountered by distant signals and wireless network provisioning play key roles in geodetic location, wh
17、ereas proximity to one or more Wi-Fi access points or Bluetooth beacons, the prevailing density of such reference points provisioned in the NEAD, and the neighboring environment are much stronger factors affecting DL and its quality. As such, factors that are to be taken into account in DL testing a
18、re, to a significant degree, different or prioritized differently from those affecting geodetic location. A modified test process distinct from x/y testing is therefore required. This Clause with its sub-clauses presents the framework of this test methodology and its relevant details without being o
19、verly prescriptive on test point selection to allow a degree of flexibility in DL test implementation. 7.1 Where to Test The methodology developed for the 9-1-1 Location technologies Test Bed in ATIS-0500031.v002 has selected the San Francisco and Atlanta regions for the Test Bed. This combination i
20、s also sufficiently diverse, from a dispatchable location perspective, to continue to be recommended as the Test Bed regions for DL testing. Urban Atlanta does not look like urban San Francisco and neither do the respective suburban or rural areas look alike. Accordingly, testing in these distinct t
21、est bed regions explores not only the effects of distinct construction methods and materials on Wi-Fi signals, but also the effects of regional differences in provisioned data, their sources, and their densities. This East-West combination of test bed regions continues to provide sufficient diversit
22、y to be a good representation of a large majority of the nation. In each region, different building types and environmental densities affect dispatchable location. This is often through the density of reference points in the area and the penetration or diffusion of Wi-Fi signals from such reference
23、points. Although Wi-Fi-based positioning is promising in the context of metric threshold-based accuracy in challenging urban environments, it can be susceptible to indicating an erroneous structure. This depends on a number of factors, including building type and size and proximity to other building
24、s with Wi-Fi transmitters, as well as the algorithms implemented in the wireless networks location server. This has to be tested thoroughly in the context of DL, including the effects of what level of provisioning in the NEAD, as well as decisions by the wireless location servers. Testing of DL ther
25、efore has to be performed in different morphologies to capture the distinct densities and predominant building use types. For example, a dense urban area typically has a high density of taller business/commercial buildings, whereas urban areas, which may have buildings of similar heights in some pla
26、ces, have a lesser density with a mix of commercial and residential buildings. The use type of the building, whether commercial or residential, affects dispatchable location in three ways: (i) the distinct density of provisioned Reference Points (RPs) between residential and commercial, depending on
27、 their available provisioning sources; (ii) ATIS-0500035 8 differences in the internal structure of the building as well as its actual internal and external construction materials; and (iii) the presence of neighboring residential or commercial buildings surrounding the target building. Not only is
28、it necessary to test in the four well-established distinct morphologies, but it is also necessary to include sufficiently diverse elements of each morphology, e.g., in a suburban morphology conduct tests in a suburban home surrounded by other suburban homes as well as in a suburban home next to one
29、or more large apartment complexes. These two example cases may not behave differently for Assisted Global Navigation Satellite System (A-GNSS) but can behave quite differently for dispatchable location. 7.2 Distribution of Test Buildings e.g., is it DL1 in the correct zone of the building and within
30、 +/-1 floor, or DL2 in the correct unit, or only a civic location; the distribution of test points in each building needs to support such determination. It is recommended that a number of points per building and per test floor larger than have been used in Stages 1 and 2 of the Test Bed be identifie
31、d as test points. (This is partially offset by significantly fewer test calls needed at each test point.) Test points on a given floor need to be as much as possible in specific units in specific zones of the building. They also need to be in both interior and exterior parts of the building. Interio
32、r and exterior parts of a floor will have different Wi-Fi environments and likely different DL behavior. A test point inside a given unit can serve as both in that unit as well as in that quadrant of the building where the unit lies for the purposes of DL1 and DL2 classification of results. For larg
33、er units, e.g., a large suite occupying half a floor or more, testing in more than one room with different surroundings should be planned as well. While testing in interior hallways or near elevators of commercial or residential buildings is acceptable for geodetic location testing, it can often lea
34、d to ambiguous results for dispatchable location. The preference therefore is for test points in specific addressable units in such buildings. However not every test point needs to be in a very specific addressable unit; occasionally a test point can be in a clearly identifiable zone or quadrant of
35、a building, e.g., main first floor hallway by rear exit (in northwest quadrant). In large, more public buildings, e.g., arena or museum, clearly identifiable segments of space or rooms in the building can be used, e.g., section 5 of the 2nd level of the seating area, employee training room, 3rd floo
36、r cafeteria, or the like. In taller buildings, testing needs to be performed in low floors, middle floors, and upper floors. Testing in at least 2 consecutive floors in each floor range is recommended. A minimum of 2 and preferably 3 test points per floor are recommended, even more for buildings wit
37、h large footprints (e.g., an arena). The range of test points per building will be from 4 in a 2-story house to approximately 20 in larger high rises. Specific recommendations for the different building types and corresponding numbers of test points are provided in Clause 7.9. It is clear that the t
38、est building and test point requirements for testing DL are distinct from those of testing geodetic location accuracy. Buildings that have been identified for geodetic location testing could be leveraged if they meet the various requirements for DL testing, regrading access and adequate availability
39、 of provisioned reference points in the NEAD. ATIS-0500035 9 7.4 NEAD Reference Point Requirements for Test Buildings In general, smaller test buildings should have one or more potential reference point records in the NEAD. If there are none, then it is known beforehand that the DL results in that b
40、uilding will likely be erroneous or null. The null case, when a building has no RPs and is far from other buildings, is not interesting and wasteful of test resources. However, the other case when a building has no provisioned RPs and is close to another that has RPs provisioned in the NEAD is an in
41、teresting case that could be encountered with some frequency. It is therefore recommended that one to two such buildings, preferably, of different sizes be included among the set of 20 buildings in each Test Bed region. At the other end of the required RP density, a commercial high rise should have
42、multiple potential records that are representative of the prevailing level of provisioning in the NEAD. What matters is that the chosen sample of buildings in a given morphology polygon be representative of similar buildings and the overall morphology in the Test Bed region. The fundamental premise
43、of testing in the Test Bed is to perform representative testing, in the sense that this testing attempts to replicate the conditions and location system performance that would prevail in other parts of the country, e.g., in the six monitoring regions, as well as other areas across the country. Testi
44、ng dispatchable location in the Test Bed is no different. The Test Bed should be viewed as a microcosm for the nation, hence no special procedures affecting performance, e.g., extra RP provisioning density, localized provisioning, calibration, etc. affecting the Test Bed areas exclusively should be
45、attempted. It should be stated, however, that regional variation, e.g., in sources of provisioning NEAD data, will exist and it is not the intent to avoid this natural variation from occurring. In each of dense urban, urban, and suburban morphologies in the two Test Bed regions it is expected that s
46、ufficient reference point records will be provisioned in the NEAD prior to the start of DL testing. It is also expected that within the polygons currently defined in each of these morphologies, as specified in ATIS-0500031.v002, there will be adequate reference point records for effective DL testing
47、. Special Considerations for the Rural Environment. 7.5 Special Considerations for the Rural Morphology In the rural Atlanta polygon it is expected that adequate records are likely to be available, although if challenges arise in securing the relatively small required number of test buildings, expan
48、sion of that polygon might be needed. As for the rural polygons of the San Francisco region per ATIS-0500031.v002, those lie in CMA 339, which is not likely to be provisioned in the NEAD in the time frame for early dispatchable location testing. It is recommended that a rural portion of southern San
49、ta Clara County, e.g., in the area between or adjacent to Morgan Hill and Gilroy, be used to define an alternate rural polygon for DL test purposes. The main reason behind selecting rural polygons further away for the San Francisco Region, i.e., in the foothills of the Sierras or the Central Valley, is a lower prevailing cell site density like many rural parts of the nation. Cell site density is a factor much more relevant to testing geodetic location and does not apply significantly to DL. Hence, testing DL in an essentially rural environment in southern Santa Clara County would be a
