1 internet. These wearable networks are based in part on emerging wireless technologies with very low power consumption such as narrow-band systems which were proposed as a part of the IEEE 802.15.6 WBAN standard [1]. A WBAN is operating in a heterogeneous environment where different wireless technologies coexist in the same frequency band. Thus, in order to insure reliable data transmission , time synchronization is a crucial task in a WBAN receiver providing robustness to noise, interference and contention. As time synchronization is processed at front-end, a bad synchronization degrades the whole receiver. Both Data Aided (DA) and Non Data Aided (NDA) time delay estimation techniques have been employed in real systems such as in [2] and [3]. Even if DA techniques lead to the best achievable performance, one needs to look for other solutions not requiring the transmission of a pilot signal in order to enhance the spectral efficiency and to save power. To deal with this problem, NDA time recovery techniques are implemented using only the received signal. Unfortunately, compared to DA techniques , the system performance is degraded especially at low Signal to Noise Ratio (SNR) values. To overcome such disadvantages, with Abstract: In this paper, we present a maximum likelihood (ML) based time synchronization algorithm for Wireless Body Area Networks (WBAN). The proposed technique takes advantage of soft information retrieved from the soft demapper for the time delay estimation. This algorithm has a low complexity and is adapted to the frame structure specified by the IEEE 802.15.6 standard [1] for the narrowband systems. S i m u l a t i o n r e s u l t s h a v e s h o w n g o o d performance which approach the theoretical mean square error limit bound represented by the Cramer Rao Bound (CRB).