Author Contributions
Conceptualization, A.A., M.H., O.A.A., R.A.R. and A.S.A.; methodology, A.A., M.H., O.A.A., R.A.R. and A.S.A.; software, A.A. and M.H.; validation, A.A., M.H., O.A.A. and R.A.R.; formal analysis, A.A., M.H. and O.A.A.; investigation, A.A., M.H., O.A.A., R.A.R. and A.S.A.; resources, A.A., M.H., O.A.A., R.A.R. and A.S.A.; data curation, A.A., M.H., O.A.A. and R.A.R.; writing—original draft preparation, A.A., M.H., O.A.A. and R.A.R.; writing—review and editing, A.A., M.H., O.A.A., R.A.R. and A.S.A.; visualization, A.A., M.H. and O.A.A.; supervision, O.A.A. and R.A.R.; project administration, O.A.A., R.A.R. and A.S.A.; funding acquisition, O.A.A., R.A.R. and A.S.A.; All authors have read and agreed to the published version of the manuscript.
Figure 1.
Illustration of the floor layout of buildings F03 and F04. All wings of floor 3 of F03 and all wings of floors 3,4 of F04 have the reported layout. The figure is not drawn to scale.
Figure 1.
Illustration of the floor layout of buildings F03 and F04. All wings of floor 3 of F03 and all wings of floors 3,4 of F04 have the reported layout. The figure is not drawn to scale.
Figure 2.
Illustration of the floor layout of buildings CX1 and CY2. All wings of floor 3,5 of CX1 have the reported layout. Floor 6, wing A, building CY2 has the same layout as wing A, floor 3/5 building CX1. The figure is not drawn to scale.
Figure 2.
Illustration of the floor layout of buildings CX1 and CY2. All wings of floor 3,5 of CX1 have the reported layout. Floor 6, wing A, building CY2 has the same layout as wing A, floor 3/5 building CX1. The figure is not drawn to scale.
Figure 3.
Floor maps of buildings CX1 and CY2, illustrating symmetry in AP deployment.
Figure 3.
Floor maps of buildings CX1 and CY2, illustrating symmetry in AP deployment.
Figure 4.
Floor maps of buildings F03 and F04, illustrating symmetry in AP deployment.
Figure 4.
Floor maps of buildings F03 and F04, illustrating symmetry in AP deployment.
Figure 5.
Screenshot of the Android application, WiFi Analyzer used for AP identification and labeling.
Figure 5.
Screenshot of the Android application, WiFi Analyzer used for AP identification and labeling.
Figure 6.
Descriptive dual-band MAC address labeling scheme.
Figure 6.
Descriptive dual-band MAC address labeling scheme.
Figure 7.
Snapshot of customized MATLAB®-based data acquisition GUI.
Figure 7.
Snapshot of customized MATLAB®-based data acquisition GUI.
Figure 8.
MATLAB® command used for Wi-Fi RSSI data acquisition.
Figure 8.
MATLAB® command used for Wi-Fi RSSI data acquisition.
Figure 9.
Illustrating the organization of raw data into subdirectories.
Figure 9.
Illustrating the organization of raw data into subdirectories.
Figure 10.
File naming convention applied to individual raw data files.
Figure 10.
File naming convention applied to individual raw data files.
Figure 11.
File naming convention applied to radio map files.
Figure 11.
File naming convention applied to radio map files.
Figure 12.
Power map representation of median RSSI at all distances from all sources within, 2.4 GHz training data, of device 1, in building CX1, at floor-wing (a) 3-A (b) 3-B (c) 5-A (d) 5-B.
Figure 12.
Power map representation of median RSSI at all distances from all sources within, 2.4 GHz training data, of device 1, in building CX1, at floor-wing (a) 3-A (b) 3-B (c) 5-A (d) 5-B.
Figure 13.
Power map representation of median RSSI at all distances from all sources within, 2.4 GHz training data, of device 2, in building CX1, at floor-wing (a) 3-A (b) 3-B (c) 5-A (d) 5-B.
Figure 13.
Power map representation of median RSSI at all distances from all sources within, 2.4 GHz training data, of device 2, in building CX1, at floor-wing (a) 3-A (b) 3-B (c) 5-A (d) 5-B.
Figure 14.
Power map representation of median RSSI at all distances from all sources within, 5 GHz training data, of device 1, in building CX1, at floor-wing (a) 3-A (b) 3-B (c) 5-A (d) 5-B.
Figure 14.
Power map representation of median RSSI at all distances from all sources within, 5 GHz training data, of device 1, in building CX1, at floor-wing (a) 3-A (b) 3-B (c) 5-A (d) 5-B.
Figure 15.
Power map representation of median RSSI at all distances from all sources within, 5 GHz training data, of device 2, in building CX1, at floor-wing (a) 3-A (b) 3-B (c) 5-A (d) 5-B.
Figure 15.
Power map representation of median RSSI at all distances from all sources within, 5 GHz training data, of device 2, in building CX1, at floor-wing (a) 3-A (b) 3-B (c) 5-A (d) 5-B.
Figure 16.
Power map representation of median RSSI at all distances from all sources within, 2.4 GHz training data, of device 1, in building F04, at floor-wing (a) 3-A (b) 3-B (c) 3-C (d) 4-A (e) 4-B (f) 4-C.
Figure 16.
Power map representation of median RSSI at all distances from all sources within, 2.4 GHz training data, of device 1, in building F04, at floor-wing (a) 3-A (b) 3-B (c) 3-C (d) 4-A (e) 4-B (f) 4-C.
Figure 17.
Power map representation of median RSSI at all distances from all sources within, 2.4 GHz training data, of device 2, in building F04, at floor-wing (a) 3-A (b) 3-B (c) 3-C (d) 4-A (e) 4-B (f) 4-C.
Figure 17.
Power map representation of median RSSI at all distances from all sources within, 2.4 GHz training data, of device 2, in building F04, at floor-wing (a) 3-A (b) 3-B (c) 3-C (d) 4-A (e) 4-B (f) 4-C.
Figure 18.
Power map representation of median RSSI at all distances from all sources within, 5 GHz training data, of device 1, in building F04, at floor-wing (a) 3-A (b) 3-B (c) 3-C (d) 4-A (e) 4-B (f) 4-C.
Figure 18.
Power map representation of median RSSI at all distances from all sources within, 5 GHz training data, of device 1, in building F04, at floor-wing (a) 3-A (b) 3-B (c) 3-C (d) 4-A (e) 4-B (f) 4-C.
Figure 19.
Power map representation of median RSSI at all distances from all sources within, 5 GHz training data, of device 2, in building F04, at floor-wing (a) 3-A (b) 3-B (c) 3-C (d) 4-A (e) 4-B (f) 4-C.
Figure 19.
Power map representation of median RSSI at all distances from all sources within, 5 GHz training data, of device 2, in building F04, at floor-wing (a) 3-A (b) 3-B (c) 3-C (d) 4-A (e) 4-B (f) 4-C.
Figure 20.
Position-wise mean RSSI values of four APs with highest RSSI in 2.4 GHz training data, of device 1, in building F04, at floor-wing (a) 3-A (b) 3-B (c) 3-C.
Figure 20.
Position-wise mean RSSI values of four APs with highest RSSI in 2.4 GHz training data, of device 1, in building F04, at floor-wing (a) 3-A (b) 3-B (c) 3-C.
Figure 21.
Position-wise mean RSSI values of four APs with highest RSSI in 2.4 GHz training data, of device 1, in building F04, at floor-wing (a) 4-A (b) 4-B (c) 4-C.
Figure 21.
Position-wise mean RSSI values of four APs with highest RSSI in 2.4 GHz training data, of device 1, in building F04, at floor-wing (a) 4-A (b) 4-B (c) 4-C.
Figure 22.
The flowchart representation of proposed SWRM algorithm.
Figure 22.
The flowchart representation of proposed SWRM algorithm.
Figure 23.
The empirical CDF of positioning errors when implementing KNN based fingerprinting algorithm, using building CX1 test data of (a) device 1 at 2.4 GHz (b) device 1 at 5 GHz (c) device 2 at 2.4 GHz (d) device 2 at 5 GHz.
Figure 23.
The empirical CDF of positioning errors when implementing KNN based fingerprinting algorithm, using building CX1 test data of (a) device 1 at 2.4 GHz (b) device 1 at 5 GHz (c) device 2 at 2.4 GHz (d) device 2 at 5 GHz.
Figure 24.
The empirical CDF of positioning errors when implementing KNN based fingerprinting algorithm, using building CX1, dual-band test data of (a) device 1 only (b) device 2 only (c) device 1 and 2.
Figure 24.
The empirical CDF of positioning errors when implementing KNN based fingerprinting algorithm, using building CX1, dual-band test data of (a) device 1 only (b) device 2 only (c) device 1 and 2.
Figure 25.
The empirical CDF of positioning errors when implementing KNN based fingerprinting algorithm, using building F04 test data of (a) device 1 at 2.4 GHz (b) device 1 at 5 GHz (c) device 2 at 2.4 GHz (d) device 2 at 5 GHz.
Figure 25.
The empirical CDF of positioning errors when implementing KNN based fingerprinting algorithm, using building F04 test data of (a) device 1 at 2.4 GHz (b) device 1 at 5 GHz (c) device 2 at 2.4 GHz (d) device 2 at 5 GHz.
Figure 26.
The empirical CDF of positioning errors when implementing KNN based fingerprinting algorithm, using building F04, dual-band test data of (a) device 1 only (b) device 2 only (c) device 1 and 2.
Figure 26.
The empirical CDF of positioning errors when implementing KNN based fingerprinting algorithm, using building F04, dual-band test data of (a) device 1 only (b) device 2 only (c) device 1 and 2.
Figure 27.
The empirical CDF of positioning errors in all corridors and open spaces within building F04.
Figure 27.
The empirical CDF of positioning errors in all corridors and open spaces within building F04.
Figure 28.
The empirical CDF of positioning errors when implementing KNN based fingerprinting algorithm, using dual-band test data of (a) building F03, device 1 (b) building F03, device 2 (c) building CY2, device 1 (d) building CY2, device 2.
Figure 28.
The empirical CDF of positioning errors when implementing KNN based fingerprinting algorithm, using dual-band test data of (a) building F03, device 1 (b) building F03, device 2 (c) building CY2, device 1 (d) building CY2, device 2.
Table 1.
Comparative summary of related open access Wi-Fi datasets.
Table 1.
Comparative summary of related open access Wi-Fi datasets.
Wi-Fi Dataset | Total Buildings | Buildings Similarity | Total Floors/Wings | Multi Device Data Acquisition | Labeled Dual-Band Data | Network Device Homogeneity | Spatial Symmetry |
---|
UJIIndoorLoc [34] | 3 | – | 13 | ✓ | – | – | – |
DSI [20] | 1 | – | 1 | – | – | – | – |
Minho [21] | 1 | – | 1 | – | – | – | – |
KTH [22] | 1 | – | 1 | – | – | – | – |
GEOTEC [35] | 1 | – | 1 | ✓ | ✓ * | – | – |
TUT1 [23] | 1 | – | 5 | ✓ | – | – | – |
TUT2 [24] | 2 | – | 7 | – | – | – | – |
UJI [36] | 1 | – | 2 | – | – | – | – |
JUIndoorLoc [37] | 1 | – | 3 | ✓ | – | – | – |
Escuela [33] | 1 | – | 2 | ✓ | ✓ | ✓ | – |
UTMInDualSymFi | 4 | ✓ | 14 | ✓ | ✓ | ✓ | ✓ |
Table 2.
Relevant open access datasets for indoor positioning and associated works.
Table 2.
Relevant open access datasets for indoor positioning and associated works.
Database | Wi-Fi Only | Hybrid with Wi-Fi | BLE | Magnetic Sensors |
---|
Works | [23,33,34,36,37,38] | [29,35,49,50,51] | [55,56,57] | [58] |
Data access | [19,20,21,22,25,26,27] | [39,43,44,45,46,47,52,53] | [59,60,61] | [62] |
Table 3.
Structural and APs related information of the four buildings, from where Wi-Fi RSSI data were acquired.
Table 3.
Structural and APs related information of the four buildings, from where Wi-Fi RSSI data were acquired.
Building | Total Floors | Wings per Floor | APs per Wing |
---|
1. F04 | 10 | 3 | 1 |
2. CX1 | 8 | 2 | 2 or 3 |
3. F03 | 10 | 3 | 1 |
4. CY2 | 8 | 2 | 2 or 3 |
Table 4.
Summary of basic data collection related information of each building.
Table 4.
Summary of basic data collection related information of each building.
Building | Floor(s) | Area Covered in Data Collection (m2) | Wings | Training Data | Test Data |
---|
1. F04 | 3, 4 | 252 | A, B, C | ✓ | ✓ |
2. CX1 | 3, 5 | 230 | A, B | ✓ | ✓ |
3. F03 | 4 | 126 | A, B, C | ✕ | ✓ |
4. CY2 | 6 | 60 | A | ✕ | ✓ |
Table 5.
Details of reference and test points within building 1 (F04).
Table 5.
Details of reference and test points within building 1 (F04).
Floor-Wing | RP Positions (Meters) | TP Positions/Distance (Meters) |
---|
3-A | 1:20 | 2.21, 4.63, 7.25, 8.64, 10.88, 12.42, 14.66, 16.73, 18.35 |
3-B | 1:20 | 2.38, 6.3, 8.57, 12.45, 15.78, 17.54, 19.36 |
3-C | 1:20 | 1.8, 3.87, 6.1, 7.6, 9.27, 13.62, 14.72, 18.37 |
4-A | 1:20 | 1.76, 3.32, 4.65, 7.4, 9.59, 11.74, 14.26, 18.83 |
4-B | 1:20 | 1.51, 2.55, 3.36, 5.75, 8.38, 11.4, 13.72, 18.62 |
4-C | 1:20 | 3.15, 5.66, 8.52, 10.32, 13.42, 15.81, 18.23 |
Table 6.
Details of reference and test points within building 2 (CX1).
Table 6.
Details of reference and test points within building 2 (CX1).
Floor-Wing | RP Position (Meters) | TP Positions/Distance (Meters) |
---|
3-A | 1:34 | 3.36, 7.63, 10.16, 15.62, 20.7, 24.25, 28.44, 30.77 |
3-B | 1:31 | 3.3, 7.57, 8.39, 10.66, 16.76, 19.47, 22.82, 25.39, 28.72, 30.35 |
5-A | 1:34 | 2.3, 7.45, 10.8, 14.2, 18.3, 20.6, 23.7, 26.4 |
5-B | 1:31 | 3.8, 7.55, 11.3, 12.7, 15.6, 16.3, 18.2, 20.85, 25.6, 29.65 |
Table 7.
Details of test points within building 3 (F03) and 4 (CY2).
Table 7.
Details of test points within building 3 (F03) and 4 (CY2).
Building | Floor-Wing | TP Positions/Distance (Meters) |
---|
3. F03 | 4-A | 1:20 |
3. F03 | 4-B | 1:20 |
3. F03 | 4-C | 1:20 |
4. CY2 | 6-A | 1:34 |
Table 8.
Details of access points identified and installed within each building.
Table 8.
Details of access points identified and installed within each building.
Building | Total APs | 2.4 GHz MACs | 5 GHz MACs |
---|
1. F04 | 25 | 25 | 25 |
2. CX1 | 38 | 38 | 38 |
3. F03 | 24 | 24 | 24 |
4. CY2 | 37 | 37 | 37 |
Table 9.
Details of octets within descriptive MACs assignment methodology, along with valid values.
Table 9.
Details of octets within descriptive MACs assignment methodology, along with valid values.
MAC Address Octet | Description | Possible Values |
---|
1 | Building number | 01, 02, 03, 04 |
2 | Floor number | 01, 02, 03, 04, 05, 06, 07, 08, 09, 10 |
3 | Wing | 0a, 0b, 0c |
4 | AP number | 01, 02, 03 |
5:6 | Band/Frequency | 02:40, 05:00 |
Table 10.
Details of hardware and software setup utilized in Wi-Fi RSSI data collection.
Table 10.
Details of hardware and software setup utilized in Wi-Fi RSSI data collection.
Item/Feature | Description/Detail |
---|
Data acquisition platform | Laptop (wireless network adapter) |
No. data recording devices | 2 laptops (Windows® 10 64-bit) |
Device (laptop) 1—Model | HP® ENVY x360 Convertible 15m-bp1xx |
Device 1—wireless adapter | Intel® Dual Band Wireless-AC 7265 |
Device (laptop) 2—Model | Dell™ Inspiron 15-3567 |
Device 2—wireless adapter | Qualcomm® Wireless-AC 7265 |
Table 11.
Data attributes and labels recorded during Wi-Fi RSSI data acquisition procedure.
Table 11.
Data attributes and labels recorded during Wi-Fi RSSI data acquisition procedure.
Label/Attribute | Label/Attribute |
---|
Timestamp | Building number |
RSSI and MACs of all valid sources | Floor number |
Captured data type | Wing identifier |
Data recording device | Distance (position) |
Table 12.
Details of total RPs/TPs and total test and training data samples recorded on each device, within each building.
Table 12.
Details of total RPs/TPs and total test and training data samples recorded on each device, within each building.
Feature | 1. F04 | 2. CX1 | 3. F03 | 4. CY2 |
---|
No. Reference Points (RPs) | 120 | 130 | – | – |
Training samples at each RP | 226 | 226 | – | – |
Total training data samples | 27,120 | 29,380 | – | – |
Test samples at each RP | 113 | 113 | – | – |
No. Test Points (TPs) | 47 | 26 | 60 | 34 |
Test samples at each TP | 150 | 150 | 150 | 150 |
Total test data samples | 20,610 | 18,590 | 9000 | 5100 |
Table 13.
A listing of all valid values of octets within the descriptive MACs, assigned to out of buildings detected dual-band sources.
Table 13.
A listing of all valid values of octets within the descriptive MACs, assigned to out of buildings detected dual-band sources.
MAC Address Octet | Possible Values |
---|
1 | 11, 22, 33, 44 |
2 | ff |
3 | ff |
4 | 01, 02, 03, 04, 05, 06, 07, 08, 09, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21 |
5:6 | 02:40, 05:00 |
Table 14.
The number of APs and frequency-wise MACs detected in each building during the Wi-Fi RSSI data recording.
Table 14.
The number of APs and frequency-wise MACs detected in each building during the Wi-Fi RSSI data recording.
Building | Detected APs | Detected MACs | 2.4 GHz MACs | 5 GHz MACs |
---|
1. F04 | 41 | 64 | 40 | 24 |
2. CX1 | 39 | 71 | 39 | 32 |
3. F03 | 38 | 62 | 38 | 24 |
4. CY2 | 51 | 71 | 51 | 20 |
Table 15.
Possible values and description of individual raw data file name fields, as illustrated in
Figure 10.
Table 15.
Possible values and description of individual raw data file name fields, as illustrated in
Figure 10.
Field in File Name | Possible Values and Description |
---|
Data type | 1 = training data; 2 = test data |
Device | 1 = device 1; 2 = device 2 |
Building | 1 = F04; 2 = CX1; 3 = F03; 4 = CY2 |
Floor | floor number within corresponding building (as in Table 4) |
Wing | 10 = A; 11 = B; 12 = C |
Distance | distance of RP/TP in corresponding building, floor, wing (as in Table 5, Table 6 and Table 7) |
Table 16.
Description of contents within each column of raw data .csv files. Valid for individual and combined data files.
Table 16.
Description of contents within each column of raw data .csv files. Valid for individual and combined data files.
Column Number | Description |
---|
1 | timestamp: year |
2 | timestamp: month |
3 | timestamp: day |
4 | timestamp: hours |
5 | timestamp: minutes |
6 | timestamp: seconds |
7 to Ntc - 6 | RSSI’s in dBm from each source (MAC); RSSI = 100: source not detected |
Ntc - 5 | data type: 1 = training; 2 = test |
Ntc - 4 | recording device: 1 = device 1; 2 = device 2 |
Ntc - 3 | building: 1 = F04; 2 = CX1; 3 = F03; 4 = CY2 |
Ntc - 2 | floor number within corresponding building (as in Table 4) |
Ntc - 1 | wing: 10 = A; 11 = B; 12 = C |
Ntc | distance of RP/TP in meters |
Table 17.
List of total MACs and data columns associated with raw data files of each building and frequency band.
Table 17.
List of total MACs and data columns associated with raw data files of each building and frequency band.
Building | Frequency/Band | Total No. Columns (Ntc) | Total MACs (M) |
---|
1. F04 | 2.4 GHz | 52 | 40 |
1. F04 | 5 GHz | 36 | 24 |
1. F04 | Dual band | 76 | 64 |
2. CX1 | 2.4 GHz | 51 | 39 |
2. CX1 | 5 GHz | 44 | 32 |
2. CX1 | Dual band | 83 | 71 |
3. F03 | 2.4 GHz | 50 | 38 |
3. F03 | 5 GHz | 36 | 24 |
3. F03 | Dual band | 74 | 62 |
4. CY2 | 2.4 GHz | 63 | 51 |
4. CY2 | 5 GHz | 32 | 20 |
4. CY2 | Dual band | 83 | 71 |
Table 18.
Floor and wing classification accuracy baseline results of the test scenarios for building CX1.
k is the number of nearest neighbors in KNN algorithm [
70].
Table 18.
Floor and wing classification accuracy baseline results of the test scenarios for building CX1.
k is the number of nearest neighbors in KNN algorithm [
70].
Scenario | Floor and Wing Classification Accuracy (%) |
---|
(Test Data) | | | |
---|
2.4 GHz, Device 1 | 100.0 | 100.0 | 100.0 |
5 GHz, Device 1 | 99.9 | 99.9 | 99.9 |
Dual-band, Device 1 | 100.0 | 100.
0 | 99.9 |
2.4 GHz, Device 2 | 99.8 | 99.9 | 99.8 |
5 GHz, Device 2 | 100.0 | 100.0 | 100.0 |
Dual-band, Device 2 | 100.0 | 100.0 | 100.0 |
Dual-band, Device 1 & 2 | 99.9 | 99.9 | 99.9 |
Table 19.
Positioning accuracy baseline results of the test scenarios for building CX1.
k is the number of nearest neighbors in KNN algorithm [
70].
Table 19.
Positioning accuracy baseline results of the test scenarios for building CX1.
k is the number of nearest neighbors in KNN algorithm [
70].
Scenario | 75th Percentile of Positioning Error (Meters) |
---|
(Test Data) | | | |
---|
2.4 GHz, Device 1 | 3.3 | 3.6 | 3.7 |
5 GHz, Device 1 | 2.0 | 2.4 | 2.6 |
Dual-band, Device 1 | 2.3 | 2.6 | 2.4 |
2.4 GHz, Device 2 | 3.5 | 3.4 | 3.7 |
5 GHz, Device 2 | 2.0 | 2.3 | 2.5 |
Dual-band, Device 2 | 2.2 | 2.2 | 2.2 |
Dual-band, Device 1 & 2 | 2.9 | 3.2 | 3.2 |
Table 20.
Floor and wing classification accuracy baseline results of the test scenarios for building F04.
k is the number of nearest neighbors in KNN algorithm [
70].
Table 20.
Floor and wing classification accuracy baseline results of the test scenarios for building F04.
k is the number of nearest neighbors in KNN algorithm [
70].
Scenario | Floor and Wing Classification Accuracy (%) |
---|
(Test Data) | | | |
---|
2.4 GHz, Device 1 | 83.9 | 82.3 | 82.5 |
5 GHz, Device 1 | 92.5 | 92.3 | 89.5 |
Dual-band, Device 1 | 93.0 | 93.7 | 93.2 |
2.4 GHz, Device 2 | 94.0 | 95.1 | 93.2 |
5 GHz, Device 2 | 98.1 | 98.1 | 98.1 |
Dual-band, Device 2 | 100.0 | 100.0 | 100.0 |
Dual-band, Device 1 & 2 | 97.5 | 97.4 | 97.5 |
Table 21.
Positioning accuracy baseline results of the test scenarios for building F04.
k is the number of nearest neighbors in KNN algorithm [
70].
Table 21.
Positioning accuracy baseline results of the test scenarios for building F04.
k is the number of nearest neighbors in KNN algorithm [
70].
Scenario | 75th Percentile of Positioning Error (Meters) |
---|
(Test Data) | | | |
---|
2.4 GHz, Device 1 | 2.0 | 2.2 | 2.6 |
5 GHz, Device 1 | 0.9 | 0.9 | 1.3 |
Dual-band, Device 1 | 1.0 | 1.1 | 1.4 |
2.4 GHz, Device 2 | 2.3 | 2.6 | 3.2 |
5 GHz, Device 2 | 1.6 | 2.0 | 2.4 |
Dual-band, Device 2 | 1.6 | 2.0 | 2.3 |
Dual-band, Device 1 & 2 | 2.2 | 2.6 | 2.5 |
Table 22.
Devicewise, floor and wing classification accuracy baseline results for buildings F03 and CY2, using dual-band test data.
k is the number of nearest neighbors in KNN algorithm [
70].
Table 22.
Devicewise, floor and wing classification accuracy baseline results for buildings F03 and CY2, using dual-band test data.
k is the number of nearest neighbors in KNN algorithm [
70].
Scenario | Floor and Wing Classification Accuracy (%) |
---|
(Test Data) | | | |
---|
F03, Device 1 | 90.9 | 91.6 | 93.2 |
F03, Device 2 | 100.0 | 100.0 | 100.0 |
CY2, Device 1 | 100.0 | 100.0 | 100.0 |
CY2, Device 2 | 100.0 | 100.0 | 100.0 |
Table 23.
Devicewise, positioning accuracy baseline results for buildings F03 and CY2, using dual-band test data.
k is the number of nearest neighbors in KNN algorithm [
70].
Table 23.
Devicewise, positioning accuracy baseline results for buildings F03 and CY2, using dual-band test data.
k is the number of nearest neighbors in KNN algorithm [
70].
Scenario | 75th Percentile of Positioning Error (Meters) |
---|
(Test Data) | | | |
---|
F03, Device 1 | 2.3 | 2.6 | 2.7 |
F03, Device 2 | 5.6 | 5.4 | 5.4 |
CY2, Device 1 | 4.3 | 4.2 | 4.4 |
CY2, Device 2 | 5.3 | 5.0 | 4.5 |
Table 24.
Computational time measured for generation of radio maps using traditional and proposed algorithm. Experiments were performed on the same work station and averaged over 1000 repetitions.
Table 24.
Computational time measured for generation of radio maps using traditional and proposed algorithm. Experiments were performed on the same work station and averaged over 1000 repetitions.
Training | Computational Time (Seconds) |
---|
Data | Mean | Median | SWRM |
---|
2.4 GHz, CX1 | 0.1401 | 1.7971 | 0.0542 |
5 GHz, CX1 | 0.1194 | 1.4428 | 0.0529 |
2.4 GHz, F04 | 0.1220 | 1.5419 | 0.0466 |
5 GHz, F04 | 0.0825 | 0.8458 | 0.0447 |
Table 25.
Positioning accuracy baseline results with various radio maps.
Table 25.
Positioning accuracy baseline results with various radio maps.
Test | 75th Percentile of Positioning Error (Meters) |
---|
Data | Mean | Median | SWRM |
---|
CX1, Device 1, 2.4 GHz | 3.9 | 4.0 | 3.3 |
CX1, Device 1, 5 GHz | 3.6 | 3.6 | 2.0 |
CX1, Device 2, 2.4 GHz | 4.3 | 4.3 | 3.5 |
CX1, Device 2, 5 GHz | 3.6 | 3.1 | 2.0 |
F04, Device 1, 2.4 GHz | 3.6 | 3.9 | 2.0 |
F04, Device 1, 5 GHz | 2.9 | 2.7 | 0.9 |
F04, Device 2, 2.4 GHz | 4.3 | 4.0 | 2.3 |
F04, Device 2, 5 GHz | 3.6 | 3.0 | 1.6 |