In the upcoming fifth generation cellular networks, a large variety of application scenarios is planned to be covered. These range from machine-to-machine (M2M) communication with low latency, low data-rate, high reliability and a massive number of devices per cell up to very high capacity applications with data rates greater than 1 Gbit/s for a small amount of users. Due to this, an improved flexibility and efficiency of the deployed communication systems are set into focus to support the severely diverging requirements. One approach to enhance the efficiency is the deployment of waveforms which are more robust against signal spreading due to dispersive communication channels. Here, the filterbank multicarrier approach has been shown to be a promising candidate to replace the state-of-the-art CP-OFDM scheme. Additionally, the waveform of current communication systems is selected to provide the best performance in a certain amount of channel environments. Here, the design of communication systems matching their present channel environment offers some potential gains, as the present-day approach utilizes the available channel capacity sub-optimally. This thesis proposes and evaluates a new approach to design mobile communication systems called channel adaptive waveforms, in which a systems waveform in terms of subcarrier spacing and prototype filter function is optimized to match the properties of the utilized communication channel. For this purpose, the effects on data and pilot symbols transmitted over doubly dispersive channels are characterized using the (cross-)ambiguity function and the statistical properties of the communication channel. Additionally, prototype filter functions for the filterbank multicarrier system are selected based on their reconstruction performance in doubly dispersive channels. Based on these results, the practical feasibility of the proposed approach is investigated for cellular networks with homogeneous system configurations per cell. Here, all users in a cell utilize the same subcarrier spacing and prototype filter function, whereby each cell is allowed to be configured differently. With the assumption of perfect channel knowledge at the receiver, the evaluation results confirm that this system design provides significant performance gains compared to a “one-fits-all” system design approach as LTE. To obtain insight into the effects of imperfect channel knowledge, a scattered pilot based channel estimation according to the least squares approach is considered. In this regard, the channel estimation error for different scattered pilot schemes proposed in literature is derived and evaluated based on analytical expressions as well as Monte-Carlo simulations. Here, the results show that for the investigated scattered pilot schemes the performance gains of channel adaptive waveforms are reduced significantly due to residual channel estimation errors. To counteract these performance degradations, some improvements for the scattered pilot schemes are discussed briefly. Based on these results, the channel adaptive waveform approach is extended to a system setup which allows for heterogeneous system configurations per cell. For this, the related channel capacity gains are investigated for both down- and uplink transmissions. In this context, the optimization problem to maximize the capacity within a cell as well as important optimization constraints are proposed and discussed. Simulation results for a selected scenario show that channel adaptive waveforms with heterogeneous system configurations per cell provide a close to optimal channel capacity. Furthermore, the analysis of down- and uplink scenarios proves that the performance of the heterogeneous approach is basically independent from the direction of transmission. In conclusion, the outcome of this thesis shows that the application of channel adaptive waveforms in communication systems is advantageous compared to the current “one-fits-all” system design approach. For applications with a high range of channel characteristics such as cellular networks, the results indicate that channel adaptive waveforms provide significant performance gains in terms of bit-error-rates and channel capacity without the need for complex receiver structures such as multi-tap equalizers. Due to the excellent localization of the prototype filter functions applied in filterbank multicarrier systems, the amount of required guard bands is very limited. Thus, it is possible to deploy cellular networks which allow user-specific subcarrier spacing and prototype filter functions within in a single cell without sacrificing the performance gains.