A Nationwide Study on Optical Analysis for Expecting HEOs to Support Ambulances

Abstract

This paper deals with actual optical data from rural as well as urban areas in a nationwide study captured with Fisheye cameras. Simultaneously data was collected (of the receiving power density) from the mobile communications satellite N-STAR. The visibility of the satellite is easily determined by checking the value of the pixels in the binarized fisheye image of its position. The process of determining the visible satellite is automatically performed. Based on the analyses of the field data measured in Japan, we are expecting HEOs (Highly inclined Elliptical Orbiters) that would reduce blockage in the extreme northern region of Wakkanai City well as in the most crowded urban area, in Tokyo Ginza. In case of HEOs operation, the elevation angle will improve from 37 with N-STAR GEO to 75 degrees. HEOs could replace 5G/Ka-band or support in rural areas where broadband circuit is not available. We are proposing combination operations with HEOs and 5G/Ka-band to solve blockage problems, because HEOs can keep line-of-sight propagation with high elevation angle for long duration. In such operations, the communications profile on the vehicle based on actual optical data will be very useful to predict blockages and to select/switch a suitable circuit.

I. INTRODUCTION

The purpose of this paper is to promote mobile andambulatory communications, especially for ambulances. This paper discusses optical data communication in urbanand suburban areas of Japan via fisheye cameras. The approach proposed would permit the recording of simultaneously captured digital data received via the S-band from a mobile communications satellite (NTTDoCoMo’s N-STAR GEO) with an onboard tracking system, based on directional and positioning data obtained by a GPS gyrocompass.

This paper proposes methods, experiments, and a model for obtaining a nationwide communication profile to helppredict the propagation of mobile communication satellites and to support ambulatory applications, including forward video transmission links. We emphasize the benefits of deploying HEOs (Figure 1) to address issues associated with obstructions to line-of-sight communications.

 

Fig. 1. Highly inclined elliptical Orbiter (HEO) to send a real-time video from ambulance.

 

II. BACKGROUND

 

2.1. Forward linkage of video transmission from moving vehicles.

Propagation effects with the LMSS (Land Mobile Satellite Service) differ from those for fixed satellite service (FSS), primarily due to the greater significance of terraineffects, including trees, buildings, hills, and mountains.

FSS generally allows discrimination againstmultipathing, shadowing, and obstructions through the use of highly directional line-of-sight antennas placed atunobstructed sites.

The third-generation (3G) mobile communicationtechnologies already introduced in developed nations address certain data transmissions (64 kbps x n) frommoving vehicles in urban areas. However, precise observations of patient conditions by clinical physiciansdepend on high-quality images of 2 Mbps and/or 10 Mbps, with compressed HDTV from ambulances en route. Certainclinical cases require high-quality images to assess the condition of patients in critical condition, such as HDTV compressed with HEVC (High Efficiency Video Coding, or H.265), transmitted from moving ambulances.

 

2.2. The 3G cellular phone

The third-generation wireless cellular phone (the 3G wireless mobile phone) is called the 3G phase 1, or Wide-band CDMA in Japan. NTT DoCoMo has already offered full service in 99% of the surface area of Japan. In the transmission of video data from an ambulance to anemergency center, the transmission capability of the upstream line (from the mobile station to the transit node) is of great importance. NTT DoCoMo offers three kinds of upstream line services. As ambulances run at a maximum 80 km/h on the street, the data transmission is to beconducted at a rate of 144 kbps by FDD (Frequency Division Duplex) via macro Node B over a maximum transmission distance of 15 km. It is difficult to conductupstream transmission at 2 Mbps from a high speed vehicle in an environment in which frequency selective fading will be generated due to multipath interference urban areas.

 

2.3. 5G/Ka-band

5G/Ka-band is expected to support significantly fastermobile broadband speeds and higher volumes of data usethan previous generations (the latter is based on the LTE, or Long-Term Evolution, standard). The ITU-R standard for wireless broadband communication will expand the functionality of mobile devices and data terminals whileunleashing the full potential of the Internet of Things. As a core fiber-through-the-air technology, 5G is expected tosupport autonomous cars in smart cities and the industrial Internet and play a key role in the future oftelecommunications. 5G/Ka-band will also prove essentialin maintaining today’s most popular mobile applications, including on-demand video, by ensuring that growing uptake and usage can be sustained.

However, 5G/Ka-band uses wideband frequency resources with 500 MHz bandwidth on 28 GHz. This meansthe frequency characteristics related to blockage and shadowing are dramatically different from the lowermicrowave frequencies. Communication distances for 5G/Ka-band will be 100-200m. Since the operating principle is primarily line-of-sight, carriers will need todeploy many repeaters and transponders across the urbanland scape.

Since HEO can cover Ka-band obstructions, we expect HEO to prove highly effective in supporting forward videotransmission from moving vehicles to emergency triagecenters.

 

III. METHODS

 

3.1. Mathematical approach to fading problem

Multipath fading is due to the constructive and destructive combination of randomly delayed, reflected, scattered, and diffracted signal components. Under the multipath environment, there are three models designed mathematically.

 

3.1.1. Nakagami-Ricean fading

The Nakagami-Ricean density distribution is given by formula (1)

\(p(x)=\frac{2 x}{M_{r, A}} \exp \left(-\frac{1+x^{2}}{M_{r, A}}\right) I_{0}\left(\frac{2 x}{M_{r, A}}\right)\)       (1)

Io : Zero-order Bessel function M_rA: average power of scattered wave component K: Rice coefficient

\(2 \sigma^{2}=M_{\mathrm{r}, \mathrm{A}}\)       (2)

\(K=\frac{1}{M_{\mathrm{r}, \mathrm{A}}}\)       (3)

This propagation environment consists of one strong direct wave (line-of-sight) and weak multiplex propagation waves.

 

3.1.2. Loo distribution

It is assumed that the received signal is affected by nonselective Rice fading with lognormal shadowing on the direct component only, while the diffuse scattered component has constant average power level. It means thata direct wave declines by lognormal distribution caused by leaves of roadside trees. In such propagation environment, a synthetic wave ingredient will serve as Loo distribution.

\(p(x)=\frac{6.930 x}{\sigma M_{\mathrm{r}, \mathrm{B}}} \int_{0}^{\infty} \frac{1}{z} \exp \left[-\frac{[20 \log z-m]^{2}}{2 \sigma^{2}}-\frac{x^{2}+z^{2}}{M_{\mathrm{r}, \mathrm{B}}}\right] I_{0}\left(\frac{2 x z}{M_{\mathrm{r}, \mathrm{B}}}\right) \mathrm{d} z\)       (4)

\(p(x)=\frac{2 x}{M_{\mathrm{r}, \mathrm{C}}} \exp \left(-\frac{x^{2}}{M_{\mathrm{r}, \mathrm{C}}}\right)\)       (5)

MrB: average power of scattered wave component

 

3.1.3. Rayleigh distribution

When a line-of-sight signals are influenced by buildings, the direct path is entirely blocked. This propagation condition is referred as deep shadowing state, a synthetic signal is based on stray multipath components which obey Rayleigh distribution.

\(p(x)=\frac{2 x}{M_{\mathrm{r}, \mathrm{C}}} \exp \left(-\frac{x^{2}}{M_{\mathrm{r}, \mathrm{C}}}\right)\)       (6)

MrC: average power of scattered wave component

\(P\left(x_{0}\right)=1-\exp \left(-\frac{x_{0}^{2}}{M_{\mathrm{r}, \mathrm{C}}}\right)\)       (7)

 

3.2. Markov chain model

Fading among channel states is a relatively slow processif the vehicle is relatively distant from buildings or trees (forexample, 5 to 10 m). This fading is determined by the distance of user movement and elevation change, each statemaintaining a certain geometric space. The change in statesis described by the Markov chain process. When a mobileterminal moves over a certain distance, it s witches randomly to a new state. Figure 2 illustrates modeling for the three-state Markov process. We observe a strongrelationship between the three propagations of the f ading channel and visibility (optical path) taken by fisheyecamera shown in Table 1.

 

Fig. 2. Markov chain of the propagation.

 

Table 1. Schema of visibility and microwave propagation

 

3.3. Analysis of the optical method

 

3.3.1. How to obtain optical data

We captured the images recorded by the video systemusing a computer, and converted them into digital motion pictures. The motion-picture files (in the AVI format) wereconverted to still images corresponding to each frame (bitmap). To enable the easy analysis of images, the colorimages were converted to monochrome binary (two-level) images (Figure 3) in accordance with a predetermined threshold.

 

Fig. 3. Conversion to binary images.

The length from the center to the first black region of the binary image gives the minimum elevation angle in the corresponding direction. On the other hand, direction from the center to the bottom of the image agrees with therunning direction. Thus, if this direction is matched with the direction indicated by the GPS compass (the running direction of the vehicle indicated by the data synchronized using the time code and the GPS timer), the running direction of the vehicle can be calculated as an absolutevalue.

 

3.3.2. Coordinate conversion model of El and Az

First, we can calculate the satellite position from the (x,y) coordinates of the two-dimensional binary images. Totransform the (El, Az) coordinates from the (x,y) coordinates, we have to model geometric correlation shown in the Figure 4. With the following equations, we cancalculate the elevation and the azimuth of the target satellite.

x,y: coordinates of the two-dimensional still image

El, Az: gimbal coordinates of the target satellite

U, V: tangent plane at the target point of the lens

R: radius of the lens

f: focal length

star mark: target satellite

 

Fig. 4. The coordinate conversion model of El and Az using a fish-eye lens.

\(\left.\begin{array}{l} {\text { Elevation }=(\pi / 2)-\beta} \\ {\text { Azimuth }=\alpha} \\ {\mathrm{y}=\mathrm{f} \mathrm{\beta}} \end{array}\right\}\)       (8)

\(\left.\begin{array}{l} {\mathrm{x}=\frac{\mathrm{R}(\mathrm{u} \mathrm{A}-\mathrm{vB}+\mathrm{m} \mathrm{R} \sin \beta \sin \alpha)}{\sqrt{\mathrm{vu}^{2}+\mathrm{v}^{2}+\mathrm{m}^{2} \mathrm{R}^{2}}}} \\ {\mathrm{y}=\frac{\mathrm{R}(\mathrm{uC}-\mathrm{vD}-\mathrm{mR} \sin \beta \cos \alpha)}{\sqrt{\mathrm{vu}^{2}+\mathrm{v}^{2}+\mathrm{m}^{2} \mathrm{R}^{2}}}} \end{array}\right\}\)       (9)

\(\left.\begin{array}{l} {A=\cos \theta \cos \alpha-\sin \theta \sin \alpha \cos \beta} \\ {B=\sin \theta \cos \alpha+\cos \theta \sin \alpha \cos \beta} \\ {C=\cos \theta \sin \alpha+\sin \theta \cos \alpha \cos \beta} \\ {D=\sin \theta \sin \alpha-\cos \theta \cos \alpha \cos \beta} \end{array}\right\}\)       (10)

 

3.4. Hardware

Described above is the concept. Actual implementation of the scheme requires mounting the satellite receiver so as to obtain the beacon signal from N-STAR; deploying a fisheye camera with graphic image recorders; using multichannel analog records for the linear data from the experimental vehicle; leveraging GPS navigation data; and using the direction of vehicle movement detected by the GPS gyro (Figure 5, 6, 7).

 

3.4.1. Fisheye camera system

We installed the fisheye camera system on the top of theroof of the vehicle. Since the lens (Nikon FC-E8, circularimage, focal length of 8mm covering 183 degree) is designed only for COOLPIX video camera, so we have to convert image with C mount adapter. It recorded with the HDTV video recorders (Panasonic DVCPRO 50, 30 flams/sec.) Also, we mounted HDTV video camera to take a front view of the vehicle.

 

3.4.2. The N-STAR Mark-a tracking system

The N-STAR Mark-a satellites will provide fixed and mobile telecommunications services for a design life of over 10 years over the Pacific Ocean. N-STAR Mark-aprovided telecommunications coverage for the Japanese Islands and its territorial waters. We used this beacon signalon 2.5010 MHz as a pilot signal, designed a 7.5 tern helicalantenna (12dBi, C=0.95; axial type) on S-band with tracking motor (max. angler speed 30deg./sec.) based on thevehicle ’s direction data obtained by GPS gyro. The C/N of the beacon signal was over 10dB at the line of sight propagation. However, in case of severe multipathenvironment at an urban area, we selected the phasecomposition with 4 helical antennas to detect fine direction of the satellite.

 

3.4.3. GPS satellite compass

We installed a high-precision Furuno GPS satellitecompass (SC-60, accurate +/- 0.6 deg.) on the roof of thevehicle to ensure super heading accuracy. This compass iscapable of identifying rapid turns of up to 45 deg./sec. The unique tri-antenna system improves accuracy and reducesthe effects of yaw, pitch, and roll.

Since this system outputs information only on thevehicle ’s heading, identifying navigation information requires a separate commercial GPS receiver.

 

3.4.4. Driving data collection

We integrated multiple channels data to the hard disk of the PC, such as the optical 3 axis gyroscope (FOG: Fiber Optical Gyro by Japan Aviation Electronic Industry), 3 axisacceleration, the pulse (velocity pulse) from a wheel, steering angle, data of a accelerator, and data of a brake. The sampling rate of the A/D converter for the FOG was 15/sec. with 12 bits. This speed is based on the characteristic of FOG for public application recommended by the COCOM (Coordinating Committee for Multilateral Export Controls) and is not the limit speed of an A/D conversion.

 

3.4.5. Integrated image for the space-time management

The position data, date and time, heading, and speed, provided by the GPS satellite compass were superimposed on the images of the front and zenith views captured by the fisheye camera.

The vehicle heading, speed, acceleration, navigation data, and date/time were recorded on the hard disk of the vehicleorientation measurement system. The time code for the digital video camera was matched to GPS time, and the video camera was synchronized with the vehicle orientationmeasurement system with reference to GPS time. The twovideo systems mutually synchronized based on the commontime code.

 

Fig. 5. The experiment vehicle.

 

Fig. 6. Configuration of image recording

 

Fig. 7. Vehicle driving data recorder.

 

3.5. Field data collection

We gathered field data based on travel over a total distance of approximately 10,000 km (Table 2, Figure 8).

In Table 2, the distances are acquired from thetachometer of the vehicle, also showed the elevation angle to N-STAR, average value of visibility to the South. It is very difficult to find out the optical feature from this average value to the South in each city. In common sensejudgment, if location was north, a value will be low, if road width was wide, a value will be high. Then, investigation and analysis were conducted in detail to reflect from thespatial city feature more, such as the northernmost town ormost crowded shopping area.

 

3.6. Results of filed data

Here we discuss the basic field data analysis in items A, B, and C; the characteristics of Japan’s northernmost city initem D; and the largest urban area in item E.

 

A. Angle of elevation and driving direction

Figure 9 shows the minimum elevation angles observed on the south side (180 degrees) during an east-to-west drive in the suburbs. In contrast, a drive in the 45-degree direction, i.e., northeast, showed virtually no obstacles. The transverse axis represents the number of frames (distance = velocity × time). The obstacles that periodically interfere with optical visibility are utility poles.

 

B. Multipath and Loo distribution

In order to confirm the Loo distribution and multipatheffect on the S-band, we drove the experiment vehicleslowly along a road near a building and recorded the angle of elevation and the voltage of the received signal from the satellite (Figure 10).

 

Table 2. Operation distance nation-wide.

 

Fig. 8. Collected field data nation wide. Fig. 9. Minimum elevation angles for obstacles observed in adown town area during a drive from east to west. Note the numerous obstacles to the south as Japanese traffic environment.

We encountered two types of time lags, A and B, between the elevation angle and the voltage of the received signal. It is reasonable to suspect reflective waves (multipathing) willaffect A more than B, and that this multipathing is due todifferences in geometric structure from building to building and/or the direction to the satellite. We accept that errors will result when applying optical measurements tomicrowaves at low frequencies like the S-band. Theseerrors will shrink with higher frequencies.

 

Fig. 10. Relationship between received voltage and blockage, timelag A: going into the blockage, time lag B: going out from the blockage of the building.

 

C. Distance of blockage

Meanwhile, obstacles appear continuously 30-40 degrees during the drive in the downtown area. These are buildings (single-family homes in the traditional Japanesestyle). We plotted the data obtained to show the cumulative frequency of obstacle lengths in Figure 11 for the down town area.

The distance of the blockage becomes 2.3 m at an angle of elevation of 60 degrees. The most common obstacles blocking optical line-of-sight are utility poles of less than 5m in height. Both in the suburbs and downtown, allobstacles other than pedestrian bridges were less than 2 min height. Thus, the spatial diversity (width of the base between the two antennas ) on the vehicle must be at least 233 cm.

 

Fig. 11. Cumulative frequency of obstacle lengths in the down town area.

 

D. The extreme northern region of Wakkanai City

Wakkanai lies in the extreme north of Japan, located just 40 km from the island of Sakhalin. The angle of elevationto N-STAR Mark-a is just 36.8 degree. Thus, weinvestigated the angle of elevation and blockages. Figure 12 shows the route in Hokkaido, including Wakkanai.

With Japanese road environments in mind, we analyzed the relationship between visibility and azimuth angle to the south. We drove the vehicle on the test courses in down town Wakkanai, including those extending in different directions at 45-degree intervals to 360 degrees. Here, we calculated the cumulative angle of elevation and visibility aspercentage values based on a test drive over a relativelynarrow road in downtown Wakkanai. Since cars proceed on the left in Japan and approaching buildings therefore lie to the left, Figure 13 can be interpreted to indicate thesignificant obstruction to the south caused by builtstructures.

 

Fig. 12. Operation of Hokaido.

 

Fig. 13. The relationship between visibility and azimuth angle to the South, Wakkanai City

 

E. Tokyo urban area: Ginza

Ginza is currently regarded as the most densely built areain Japan. The angle of elevation to N-STAR Mark-a is 48.2 degrees. Ginza is characterized by numerous narrow alleysconnected to main streets (in red or blue; Figure 14)

 

Fig. 14. Map of Ginza, densely urbanized Tokyo neighborhood characterized by many narrow alleys connected to main streets

Here, we report on an analysis of visibility to the south (azimuth 180 degrees). Figure 15 gives overall averages, Figure 16 data for alleys, and Figure 17 data for main streets. In each figure, black represents actual field data, dark gray is urban, and light gray is suburban simulated by Prof. Yoshio Karasawa’s Formula (11) published in IEEE Trans. VT[06] .

\(\text{Visibility}=1-\text{a}(90-\text{EI})^2\\\text{EL:elevation angle[degree]>10 deg}\\\text{a}:1.43\times 10^{-4}:\text{urban}\\\text{a}: 6.0\times 10^{-5}:\text{suburban}\)(11)

Based on this result, the curve from 20 to 50 degrees is almost the same as Karasawa’s expectation, and the alleysin Ginza have severe blockages from 10 to 50 degrees. For these reasons, we anticipate HEOs will improveperformance near these obstructions and enhance 5Gcommunication for moving platforms in Japan’s urbancenters.

 

Fig. 15. The average of the alleyway and main street.

 

Fig. 16. Alleyway in Ginza.

 

Fig. 17. Main street in Ginza.

 

IV. CONSIDERRATIONS

 

4.1. Distance of the diversity antenna

Optical measurement can be a highly effective way todetect obstacles in urban environments, assuming the method presented here is applied along with a method forrecognizing differences based on the blockage of microwave propagation. All obstacles other thanoverpasses and tunnels do not exceed 2.3 m in distance.

Thus, we determined the optimal spatial diversity forambulatory applications to be 2.3 m. Prof. Kitano haspointed out the benefits of using a diversity antenna to track two satellites (GEO) published IEEE Trans. AES [07].

 

Fig. 18. Diversity distance of 2.3 m for the prototype of the n ext-generation ambulance.

 

4.2. Is it stable at an inclination of 40 degrees?

Although Russia’s MOLNIYA orbit (12 hours) or TUNDRA orbit (24 hours) have provided practical service for years in high-latitude areas, the inclination anglesinvolved are almost 63.4 degrees. The Earth is not a perfectsphere; its oblate shape generates HEO anomalies. The formula for these fluctuations is called J2; assuming aminimum, we choose an angle of inclination of 63.4 degrees (Figures 19 and 20).

 

Fig. 19. External force applied to HEOs.

 

Fig. 20. Apsidal rotation rate due to oblateness vs. inclination for values of average altitude. 63.4 degrees is the idealinclination for HEOs.

The TUNDRA orbit features significant eccentricity and an extended ellipse. Since the perigee is lower, the Tundraorbit must cross the Van Allen belt twice, resulting indegradations of solar array panels. Thus, we seek a lowereccentricity allowing passage over the Van Allen belt evenat perigee. Figure 21 shows the orbital simulation calculated with MATLAB Oribitus ED. In this case, the height of the perigee is 23,137 km, apogee 42,164 km, inclination 40 degrees, and eccentricity 0.3. This orbitwould serve all of Japan. The J2 factor of external force from the oblateness is improved (1/1,000 times smaller) than the original TUNDRA orbiter (inclination of 63.4 degrees). The calculated service duration over the angle of elevation 70 degree would cover Wakkanai for 450 minutes and Tokyo, Nagoya, and Osaka for 540 min./one(1) HEO(Figure 22). The minimum elevation angle with four (4) HEOs system shown in Figure 23. For Wakkanai, the figure improves from 37 with N-STAR GEO to 75 degrees.

 

Fig. 21. Expected HEOs for Ambulatory applications.

 

Fig. 22. Service duration over visibility angle 70 degree

Assuming four HEOs launched with Constellationshifted 6 hours, an angle of elevation by the average value of 24 hours of Tokyo, Nagoya, and Osaka will exceed 80 degrees.

 

4.3. Combination operation with HEOs and 5 G/Ka- band

Concerning about forward linkage of video transmission from moving vehicles to a center, this operation condition must be connected a broadband circuit continuously suchas 5G/Ka-band and/or HEOs. 5G/Ka-band uses 28 GHz. This means the frequency characteristics related toblockage and shadowing are dramatically different from the lower microwave frequencies. Communication distances for 5G/Ka-band will be 100-200m. Since the operating principle is primarily line-of-sight, carriers will need todeploy many repeaters and transponders across the urbanland scape. In case of 5G/Ka-band, it is predicted that shortbreak in the frame level occurs through repeaters and transponders. We are expecting combination operations with HEOs and 5G/Ka-band to solve this problem, because HEOs can keep line-of-sight propagation with highelevation angle for long duration. In such operations, the communications profile on the vehicle based on actualoptical data will be very useful to predict blockages and toselect/switch a suitable circuit.

 

Fig. 23. Minimum elevation angle with four (4) HEOs system designed by Dr. Toshihide Maeda, Hitachi Co. Ltd.

 

V. CONCLUSIONS

Visibility to a satellite is easily determined by checking the value of the pixels in the binarized fisheye image of its position. The process of determining visible satellites isperformed automatically. Although surely there is adifference from an optical course and microwave path by multipass (reflection) and/or Loo distribution, our fieldresults are nearly identical to those given by Karasawa’sformula. The launch and introduction of HEOs would significantly alleviate issues related to obstructions in urbanareas in the north of Japan (e.g., Wakkanai) and denselyurbanized areas like Ginza in Tokyo.

Here we suggest that HEOs could replace 5G in ruralareas where 5G is unavailable. In addition, we show that mobile communication profiles based on actual optical datacollection can be useful in predicting line-of-sight propagation and/or signal losses for 5G service.