A.Khaleghi, G. El Zein, I.Naqvi Institut d’Electronique et de Télécommunications de Rennes (IETR) Ali.khaleghi@insa-rennes.fr, Ghais.el-zein@insa-rennes.fr, Ijaz-haider.naqvi@ens.insa-rennes.fr

Abstract- Using time domain measurements, we assess the feasibility of time-reversal technique in ultra-wideband (UWB) communication. A typical indoor propagation channel is selected for the exploration. The channel response between receive and transmit antenna pairs is measured using time domain equipments which include an arbitrary wave generator (AWG) and a digital storage oscilloscope (DSO). The time-reversed version of the channel response is constructed with AWG and re-transmitted in the channel. The equivalent time reversed channel response is recorded. The properties of the time reversal technique in the line of sight (LOS) co-polar and cross-polar scenarios are measured.

Keywords: Time reversal, UWB, Indoor channel

1. Introduction

Time-reversal (TR) has been proposed and studied for acoustical imaging, electromagnetic imaging, underwater acoustic communication, and very recently for wireless communication [1-6]. TR is a technique to focus spatially and compress temporally broadband signals in rich scattering environment. In this technique, the time reversed version of the channel response between transmitter and receiver is used as a pre-filter for data transmission. This has several advantages for UWB communication. The equivalent TR channel response is compressed in time and has a very short effective length, thus, the complex task of estimating a large number of taps at the receiver is greatly reduced. This implies low cost receivers. Furthermore, due to the considerable focusing gain, better signal to noise ratio or equally higher data rate can be achieved. It is also possible to increase the communication range by keeping FCC spectral mask limit for UWB communication. Another feature of TR is spatial focusing that can reduce greatly the co- channel interference in multi-cell systems. The feasibility of TR in UWB has been demonstrated in [5]. However, the strategy involves indirect measurement of UWB channel in the frequency domain and then the data analysis in the time domain. In [6] the channel sounding is conducted using time- domain measurement but TR performance is demonstrated by the post-processing of the channel response. Both these approaches, give an ideal prediction of TR performance. For the first time in this context, we have acquired not only UWB channel response in time-domain using a sub-nanosecond impulse and a high performance DSO, but also we have generated the time-reversed version of the channel response using an AWG. Time reversal is activated in two LOS channels in IETR laboratory environment. The characteristics of the equivalent TR channel (focusing gain, temporal compression and temporal sidelobes) are investigated.

2. Time-reversal

Considering the transmitter-receiver pair, TR uses time reversed response of the channel as the transmitter pre-filter. Denote the channel impulse response by h(r0,τ), where r0 is the receiver location and τ is the delay variable. By applying TR technique, the effective channel response to any location r0 is thus given by

) ( ) , ( ) , ( ) , ( 1 0 0 0 τ τ τ τ n r h r X r S tr + ∗

(1)

where * denotes the convolution operation, n1(τ) is the noise component at the TR channel and Xtr is given by

( ) ( ) ) ( , ˆ , 2 0 0 τ τ τ n r h A r X tr + −

(2)

Office Office Office Corridor Metallic Chamber 16m 10m TX RX

Fig.1 Measurement environment layout

where ( ) τ − , ˆ 0r h is the time reversed version of the measured channel response that is truncated and then resampled in order constructed with AWG. A is the amplification term to supply constant transmitted power (Po) and n2(τ) is the noise components of the measured channel. We assume the noise signal for each path component as an independent additive Gaussian variants with zero mean and standard deviation σ. Thus, the influence of the noise can be alleviated by a simple averaging process over multiple measures of the channel response. Therefore, the equivalent TR channel response (1) is expressed as

( ) τ τ τ τ τ τ 1 1 0 0 0 ) ( ) ( ) , ( ) , (ˆ ) , ( n R A n r h r h A r S hh + ≈ + ∗ −

(3)

where ) (τ hh R is the autocorrelation function of the channel response that peaks at τ =0. The amount of the signal peak depends on the energy of the truncated signal. We define the focusing gain (F.G) as the ratio of the strongest tap power in TR channel to the strongest tap power of the direct channel at location r0

2 0 2 0 ) , ( max ) , ( max

…(Full text truncated)…