An MTM of a small scientific satellite and a shield box with the same size were used for this measurement in Figure 1. The MTM was a rectangular parallelepiped, 430 mm long × 470 mm wide × 435 mm high, enclosed in flat honeycomb panels made of aluminum, excluding its solar panels [12] . Its interior was divided into two almost equal subspaces by a conductive partition having many openings, and filled with mission subsystems. Although no electric or electronic components were equipped, conductive boxes and cable assemblies inside the MTM were exactly the same as the actual spacecraft. The transmitting and receiving antennas were omnidirectional, vertically polarized, low voltage-standing-wave-ratio UWB monopole antennas [13] . Their circular ground planes were 100 mm in diameter. Within the MTM, the transmitting antenna was fixed in one subspace and the receiving antenna was scanned in another, as shown in Figure 1. This position was defined the origin of the Cartesian coordinate, whose x, y, and z axes were parallel to the sides of MTM. The receiving antenna was scanned within a region 200 ≤ x [mm] ≤ 300 and −140 ≤ y [mm] ≤ 140 in 5-mm intervals on a styrofoam stage (transparent to electromagnetic waves). The feeding coaxial cables penetrated through the observation windows on the sides. During the measurements, the conductive top lid was closed.

Frequency- and time-domain propagation gains were measured with a microwave vector network analyzer (VNA). Major specifications of the measurements are listed in Table 1. The UWB propagation gains were calculated by integrating the power of the losses between the feeding points of the antennas over occupied bandwidth, as follows:

P L d B = 10 log ( 1 f H − f L ) ∫ f L f H 10 P L d B ( f ) 10 d f [ dB ] (1)

The continuous wave (CW) propagation gains at the center frequency (=6.85 GHz) were extracted therefrom. Root-mean-square delay spread was calculated from the power delay profile P(τ_i), where τ_i is the i-th path delay. Delay spread S is then defined by the square root of the second central moment of the delay profile:

S = ∑ ​ τ i 2 P ( τ i ) ∑ ​ P ( τ i ) − ( ∑ ​ τ i P ( τ i ) ∑ ​ P ( τ i ) ) 2 (2)

where the summations are taken above a threshold level −20 dB below the maximum of P(τ_i) [9] .

WiMedia [13] was used for the experiments of UWB technology to facilitate a high data rate up to 480 Mbps for wireless USB 2.0 applications. Note that maximum of 212 Mbps per node attained with SpaceWire [14] , equaling the standards of a wired onboard data bus. Its major specifications are listed in Table 2. The WiMedia, a high-speed wireless personal area communication standard, utilizes a multiband quadrature phase shift keying-orthogonal frequency division multiplexing (QPSK-OFDM) bundling up to 14 subbands each of which occupies 528 MHz. The cyclic prefix (prefixing of a symbol) used in OFDM provides a guard interval to eliminate intersymbol interference between the OFDM symbols. The multiband-OFDM employs 60.61-ns zero postfix. When the delay spreads are sufficiently shorter than 60.61 ns, therefore, the WiMedia devices can be used, which yield the maximum data rate of 480 Mbps, within the spacecrafts. Nonetheless, nominal bit rate of the device under test in this paper was 72 Mbps according to specifications.

The spatial distributions of propagation gains for CW, 10-MHz, and 7.5-GHz (the full band UWB) bandwidths within the MTM are shown in Figure 2. The UWB propagation gains did not exhibit explicit dependence on distance, whereas the IEEE 802.15.3a channel model applicable to indoor environments used the path loss increasing with the square of distance. While CW resulted in up to 28 dB fading at several “dead spots” caused by multipath interference, UWB practically yielded no dead spots. While narrowband wireless systems suffer flat fading in the multipath environments, and thus need a substantial amount of fading margin, UWB systems inherently possess anti-multipath capabilities. The cumulative distribution functions (CDFs) of propagation gains within MTM are illustrated in Figure 3. A bandwidth over 400 MHz was capable of reducing the fading depth to approximately 2 dB for the MTM [10] .

The CDFs of the delay spreads derived from measured delay profiles for full-, low-, and high-band UWB, as shown in Figure 4. The CDFs of the throughputs measured with the low- and the high-band UWB WiMedia devices are plotted in Figure 5. Apparently line-of-sight (LOS) resulted in higher throughputs than non-LOS (NLOS), while no significant differences were observed between the low- and the high-band UWBs. Various combinations of antenna, setting and polarization configuration were examined, within the same MTM. Polarization configurations were found to produce almost no effect on average power delay profiles and substantially small effects on the throughputs [11] . Matsushita et al. [15] also revealed that numerical transmission simulation resulted in sufficiently low bit error rates and high throughputs with use of time-domain propagation data measured in the MTM.

The shield box was employed to assess the whole inner space. The box had three 200 mm × 200 mm square holes: one in the center of the top lid and the others in the center of both side surfaces, as partly appeared in Figure 6. For changing total area of apertures, conductive 240 mm × 240 mm square panels with or without circular apertures were attached on the holes with use of conductive gaskets between mating surfaces. The diameter of the circular apertures was 12.5, 25, 50, 100, or 200 mm. Within the shield box, the transmitting antenna was fixed at 250 mm above the bottom and adjacent to the center of a 430-mm side, as shown in Figure 7. The receiving antenna was scanned within a region 0 ≤ x [mm] ≤ 340 and −140 ≤ y [mm] ≤ 140 in 20-mm intervals on a polystyrene foam stage. Since the minimum distance between the antenna electric centers was 120 mm < λ_L/2π (λ_L is the wavelength at the lowest frequency), all the measurements were carried out in a far-field region.

Without apertures: The spatial distributions of CW and UWB propagation gains within the shield box are depicts in Figure 8, where no data exists in the vicinity of the origin (the transmitting antenna) [10] . While CW resulted in up to 35 dB fading at several “dead spots” caused by multipath interference, propagation gain variations were 4.7, and 4 dB for full-, low-, and high-band UWB, respectively. Cumulative distribution functions (CDFs) of the UWB propagation gain are plotted in Figure 9. The fading depths versus frequency bandwidth at the deepest dead spot in the shield box were derived from measured data, as shown in Figure 10, where the center frequency was fixed at 6.85 GHz, and the 7.5-GHz-bandwidth (from 3.1 to 10.6 GHz) propagation gain was set to the 0-dB reference. A bandwidth over 400 MHz (a fractional bandwidth over 6%) was capable of reducing the fading depth to approximately 2 dB for the shield box. The UWB propagation gains did not exhibit explicit dependence on distance, whereas the IEEE 802.15.4a personal area network channel model [16] applicable to indoor environments used the path loss increasing with the square of distance. CDFs of the delay spreads derived from measured delay profiles within the shield box for full-, low-, and high-band UWB, as shown in Figure 11. Under these con ditions, the WiMedia device cannot be used, since delay spread exceeds the guard interval length.

With apertures: The UWB propagation gains were almost invariable for the total area of apertures between 10⁻⁴ and 10⁻³ m² or the relative area of the apertures between 0.01% and 0.1%, and gradually decreased with the area beyond 10⁻³ m² or 0.1%, as shown in Figure 12. The UWB propagation gain with apertures was found approximately −7.5 dB lower than that without them in average. This was attributed to radiofrequency power leakage through the apertures to the outside. The lower UWB propagation gains in the high-band UWB were ascribable to larger free space propagation losses. CDFs of the UWB propagation gain in the box with apertures are plotted in Figure 13. Medians (at which CDF = 50%) were found with apertures:

−15.9 dB for full-band UWB,

−15.2 dB for low-band UWB, and

−17.8 dB for high-band UWB.

The UWB propagation gains near the apertures did not exhibit different behavior from those in the other regions.

The delay spreads were found at less than 10 ns for the shield box with apertures, and more than 45 ns without aperture. The conductive shield box without apertures caused abundant multipath. The long delays were suppressed with apertures. Spatial distributions of the delay spread with apertures are shown in Figure 14 with apertures (total area = 9.4 × 10⁻² m²). The delay spreads were gradually decreased, as shown in Figure 15. The delay profiles and the delay spreads were found statistically invariable between the regions near and far from the apertures.

The link throughputs were measured with use of a pair of WiMedia devices, one of which was links to a solid state drive (SSD) via USB 2.0 interface, and the other was connected to a personal computer with a built-in SSD via a PCMCIA interface. Since the throughputs fluctuated typically ±4 Mb/s per trial, a number of trials (normally 35) were carried out to reduce the variation within ±1 Mb/s. The throughputs measured at (300, 0) were limited approximately beneath 72 Mbps below the total aperture area of 4.0 × 10⁻³ m², and gradually increased with the area beyond 3 × 10⁻³ m², as shown in Figure 16.