Experimental evaluation of UWB wireless communication within PC case J. Gelabert, D.J. Edwards and C.J. Stevens A novel concept of the use of an ultra-wide bandwidth (UWB) ultra- small-scale wireless interconnect scheme inside an electrically small enclosure is proposed. The concept is presented using a PC tower case as a model environment in which a channel measurement cam- paign has been made with two different types of antennas. Ricean K factor analysis was conducted for different positions of the antennas to identify the multipath behaviour of the channel, as well as channel capacity. An average maximum channel capacity of almost 6 Gbit/s is achieved using the whole UWB band for one of the antennas, assum- ing a signal-to-noise ratio of 25 dB. This work supports the proposal that a high-capacity UWB wireless interconnect scheme for communi- cating different devices within a PC case scenario may be used to replace wired interconnections, providing a wireless backplane system. Introduction: Look inside a conventional PC enclosure and one sees a big range of cables and line buses connecting different components. Part of these cables connects the main board (MB) to devices such as CD-ROM or floppy disk. These wired buses have a maximum data rate of no more than 3 Gbit/s using the serial advanced technology attachment (SATA II) with future improvements [1]. If we look closer we can find another type of thin cable integrated in different boards, such as the MB or peripheral cards (PCBs), interconnecting different microchips of their own substrate, where the standard data rate for a 32-bit communication will not exceed 1 Gbit/s [2]. All these intercon- nections require a huge amount of space, where further improvements of data rate would be limited by this fixed architecture. Replacing part of these wired cables inside small cavities has become an important future technology receiving increasing research interest [3]. We propose in this study to substitute the wired bus connections by small antennas, in order to enable a wireless interconnect between com- ponents. This study is based in the ultra-wide band (UWB, 3– 11 GHz) [4]. However we ignore FCC power regulations owing to the considerable effort expended in the electromagnetic compatibility field in ensuring that equipment enclosures do not permit radiation to leak out. The use of the UWB band is vital to achieve the highest data rates possible to match SATA II or PCI bus technologies. DVD case floppy PSU y-axis x-axis RX-floppy port 2 port 1 PC GPIB z-axis MB VNA 15 cm 30 cm 44.15 cm 20 cm 42cm 9 cm 14 cm 16 cm 12 cm 15 cm 19 cm 15 cm 13 cm 6 cm x -y positioning arm broadband printed bow-tie antenna in horizontal position PC tower case PCI card 2 RX-PCB PCI card 1 Fig. 1 Schematic diagram of environment VNA: vector network analyser; GPIB: general purpose interface bus; PSU: power supply unit; MB: main board An experimental study of a very small-scale (≤10 wavelengths) UWB channel characterisation in a shielded PC tower case environment is pre- sented. The measurement system is based upon the classic vector network analyser (VNA) technique [5]. Two identical broadband printed bow-tie antennas and Vivaldi antennas are used as a transmitter (TX) and receiver (RX), respectively. The RX antenna is fixed for all measurements in only two locations (RX-floppy and RX-PCB), whereas the TX antenna adopts different positions in an established grid. Ricean K factor and channel capacity are evaluated to identify the optimum performance of the channel. Experimental setup: An UWB propagation measurement was con- ducted using UWB signals on a rectangular aluminium PC tower case, the structure of which is shown in Fig. 1. A motor driven x-y positing arm was utilised to move around the TX in between both PCI cards. An area of 99 cm 2 (34 × 11 points separated by 5 mm each) is covered in the x-plane. The VNA was used to measure the S 21 scatter parameters in each position. The receiver was located in two fixed points as shown in Fig. 1, where RX-floppy represents the receiver antenna located next to the floppy disk area and RX-PCB is the one situ- ated just below the edge of the PCI1 card. Both would represent possible locations where a receiver antenna would be located to interconnect devices inside the computer. Experimental results and analysis: Each recorded S 21 was analysed to determine key channel parameters, including the Ricean K factor (the ratio of the power received in the line-of-sight component, first multi- path, to the total power received via indirect scattered paths, all other multipaths [6]). This study aims to show how much transmission is related to the indirect multipaths, especially due to the highly reflective and scattering environment that we have. From inspection of Fig. 2, the highest average K factor occurs when the RX is located at the PCB and has an average of 26.5 dB (22% of the received energy takes the direct path). For the other three cases mean values are all between 29 and 210 dB; only 10% of the received energy here is line of sight. For the majority of these data (.80% of all measurements) the channel is dominated by non-line-of-sight (NLOS) transmission. RX-floppy is located in a slight better line-of-sight (LOS) position than the RX- PCB, in reference to our transmitter locations, which might be shadowed by the PCB cards themselves. Even then, it is easy to conclude that the nature of our channel is mainly NLOS, where only 20% of the locations of the TX when our receiver is located in RX-floppy presents a situation where the LOS nature is higher than NLOS. These locations mainly rep- resent the positions where the TX is closer to the edge of the PCI card, where PCI2 does not interfere directly with the sight between both antennas. This study also implies that very directional antennas, like Vivaldi ones, would not be the indicated ones to be used under these conditions, where the LOS path might be partially or completely obstructed by different devices. –14 –12 –10 –8 –6 –4 –2 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 K factor probability empirical CDF bow-tie at RX-floppy Vivaldi at RX-floppy bow-tie at RX-PCB Vivaldi at RX-PCB Fig. 2 CDF of Ricean K factor for combined data for both types of antenna at RX-floppy and RX-PCB 50 55 60 bow-tie at RX-floppy Vivaldi at RX-floppy bow-tie at RX-PCB Vivaldi at RX-PCB 5 10 15 20 0 1 2 3 4 5 6 7 number of maxima multipaths channel capacity (× 10 9 ) SATA II PCI bus Fig. 3 Experimental channel capacity for combined data for both types of antenna at RX-floppy and RX-PCB The second analysis taken into account for this Letter is the Shannon capacity, which is the ultimate limit to the amount of data [7] that may be transmitted over the communication channel inside the PC tower case. The calculated channel capacity was computed for the whole UWB ELECTRONICS LETTERS 23rd June 2011 Vol. 47 No. 13