This dissertation is focused on the research efforts to develop the growth, processing, and modeling technologies for GaN-based Heterojunction Field Effect Transistors (HFETs). The interest in investigating GaN HFETs is motivated by the advantageous material properties of nitride semiconductor such as large band gap, large breakdown voltage, and high saturation velocity, which make it very promising for the high power and microwave applications. Although enormous progress has been made on GaN transistors in the past decades, the technologies for nitride transistors are still not mature, especially concerning the reliability and stability of the device. In order to improve the device performance, we first optimized the growth and fabrication procedures for the conventional AlGaN barrier HFET, on which high carrier mobility and sheet density were achieved. Second, the AlInN barrier HFET was successfully processed, with which we obtained improved I-V characteristics compared with conventional structure. The lattice-matched AlInN barrier is beneficial in the removal of strain, which leads to better carrier transport characteristics. Furthermore, new device structures have been examined, including recess-gate HFET with n+ GaN cap layer and gate-on-insulator HFET, among which the insertion of gate dielectrics helps to leverage both DC and microwave performances. In order to depict the microwave behavior of the HFET, small signal modeling approaches were used to extract the extrinsic and intrinsic parameters of the device. An 18-element equivalent circuit model for GaN HFET has been proposed, from which various extraction methods have been tested. Combining the advantages from the cold-FET measurements and hot-FET optimizations, a hybrid extraction method has been developed, in which the parasitic capacitances were attained from the cold pinch-off measurements while the rest of the parameters from the optimization routine. Small simulation error can be achieved by this method over various bias conditions, demonstrating its capability for the circuit level design applications for GaN HFET. Device physics modeling, on the other hand, can help us to reveal the underlying physics for the device to operate. With the development of quantum drift-diffusion modeling, the self-consistent solution to the Schrödinger-Poisson equations and carrier transport equations were fulfilled. Lots of useful information such as band diagram, potential profile, and carrier distribution can be retrieved. The calculated results were validated with experiments, especially on the AlInN layer structures after considering the influence from the parasitic Ga-rich layer on top of the spacer. Two dimensional cross-section simulation shows that the peak of electrical field locates at the gate edge towards the drain, and of different kinds of structures the device with gate field-plate was found to efficiently reduce the possibility of breakdown failure.