Future-generation wireless and mobile networks are expected to support a panoply of multimedia services, ranging from voice to video data. There is also a de facto "anytime anywhere" mentality that reliable communications should be ubiquitously guaranteed, irrespective of temporal or geographical constraints. However, the implicit catch is that these specifications should be achieved with only minimal infrastructure expansion or cost increases. In this thesis, various signal processing methodologies conducive to attaining these goals are presented.

First, a system model that takes into account the time-varying nature of the mobile environment is developed. To this end, a mathematically tractable basis-expansion model (BEM) of the communication channel, augmented with multiple-state characterization, is proposed. In the context of the developed system model, strategies for enhancing the quality of service (QoS), while maintaining resource efficiency, are then studied. Specifically, dynamic channel tracking, adaptive modulation and coding, interpolation and random sampling, and spatiotemporal processing are examined as enabling solutions. Next, the question of how to appropriately aggregate these disparate methods is recast as a nonlinear constrained optimization problem. This enables the construction of a flexible framework that can accommodate a wide range of applications, to deliver practical network designs. In particular, the developed methods are well-suited for multi-user communication systems, implemented using spread-spectrum and multi-carrier solutions, such as code division multiple access (CDMA) and orthogonal frequency division multiplexing (OFDM).

Moreover, privacy and security requirements are increasingly becoming essential aspects of the QoS paradigm in communications. Combined with the advent of novel security technologies, such as biometrics, the conventional communication infrastructure is expected to undergo fundamental modifications to support these new system components and modalities. Therefore, within the same framework for maximizing resource efficiency, several unique signal processing applications in network security using biometrics are also investigated in this thesis. It is shown that a resource allocation approach is equally appropriate, and productive, in delivering efficient and practical key distribution and biometric encryption solutions for secure communications.