This thesis deals with the localization and communication by means of radio signals. Based on the time-of-arrival (ToA), the first part of this thesis addresses the estimation of the position of a mobile station and the enhancement of the estimation accuracy. The second part of this thesis evaluates the channel capacity of the multiantenna ultrawideband communication channels in order to predict the achievable data rate.In the first part, a classical ToA estimation approach is considered in conjunction with a path loss model. It is shown that the path loss exploitation increases the ToA estimation accuracy, provided that the path loss exponent (PLE) is perfectly known at the receiver. Apart from such prospect performance improvement, the performance of the geolocalization by exploiting the path attenuation information is also investigated when the PLE is not exactly known. It is found that for a small deterministic PLE error and a low signal-to-noise ratio, the maximum likelihood (ML) estimator can still benefit from the path loss model even when the the path loss model is subject to the uncertain PLE. More precisely, the ML estimator under the imperfect PLE can provide lower root mean square error than the conventional maximum correlation estimator, which does not require any PLE knowledge. The above results, which are obtained for real-valued data models, is then extended to complex-valued data models. It is found that a special complex data model outperforms the real data model by a factor 1/4 in terms of the asymptotic error variance. Applying the ToA estimation with the path gain attenuation information to the wireless non-line-of-sight (NLoS) geolocation problem, we analyze the Cramér-Rao bound of the mobile position estimation.In the second part, the channel capacity is derived for a standard channel including cluster feature, which is further extended to the multiantenna systems taking into account the propagation delay across antenna elements. For uncorrelated fading, the upper bounds and the approximate channel capacity using the determinant approximation do not depend on the antenna element time delay. The channel capacity increases with the number of rays. However, it will be saturated when the number of rays becomes sufficiently large. In addition, the channel capacity of the standard NLoS channel models is higher than that of the standard line-of-sight channel models.