Next-generation wireless systems will provide services requiring accurate position information of mobile users, as e.g. in vehicular and industrial communication scenarios. Accuracy in the order of tens of centimeters will be required, which cannot be provided by standard radio positioning techniques used in legacy systems. This work proposes a method for combining multiple radio signals, e.g. in aggregated carrier components or multiple radio interfaces, in such way that reference symbols transmitted therein can be used for a joint estimation of radio signals' time of arrival. Analysis of the Cramer-Rao Lower Bound and its numerical evaluation using the Maximum-Likelihood estimator clearly show that, also depending on the total signal bandwidth and reference symbols' density, performance gains of above 10 dB can be achieved. Moreover, signal processing and practical receiver architectures are described, which consider tradeoffs between computational complexity, reference symbols overhead and performance gains.