The Tokamak à Configuration Variable (TCV) is a medium size magnetic confinement thermonuclear fusion experiment designed for the study of the plasma performances as a function of its shape. It is equipped with a high power and highly flexible electron cyclotron heating (ECH) and current drive (ECCD) system. Up to 3 MW of 2nd harmonic EC power in ordinary (O2) or extraordinary (X2) polarization can be injected from TCV low-field side via six independently steerable launchers. In addition, up to 1:5 MW of 3rd harmonic EC power (X3) can be launched along the EC resonance from the top of TCV vacuum vessel. At high density, standard ECH and ECCD are prevented by the appearance of a cutoff layer screening the access to the EC resonance at the plasma center. As a consequence, less than 50% of TCV density operational domain is accessible to X2 and X3 ECH. The electron Bernstein waves (EBW) have been proposed to overcome this limitation. EBW is an electrostatic mode propagating beyond the plasma cutoff without upper density limit. Since it cannot propagate in vacuum, it has to be excited by mode conversion of EC waves in the plasma. Efficient electron Bernstein waves heating (EBH) and current drive (EBCD) were previously performed in several fusion devices, in particular in the W7-AS stellarator and in the MAST spherical tokamak. In TCV, the conditions for an efficient O-X-B mode conversion (i.e. a steep density gradient at the O2 plasma cutoff) are met at the edge of high confinement (H-mode) plasmas characterized by the appearance of a pedestal in the electron temperature and density profiles. TCV experiments have demonstrated the first EBW coupling to overdense plasmas in a medium aspect-ratio tokamak via O-X-B mode conversion. This thesis work focuses on several aspects of ECH and EBH in low and high density plasmas. Firstly, the experimental optimum angle for the O-X-B mode conversion is successfully compared to the full-wave mode conversion calculation of the AMR code. The implementation of TCV ECH system geometry in AMR and the coupling of AMR to the LUKE quasi-linear Fokker-Planck solver for the TCV environment were part of this work. The power deposition location of modulated EBH is then detected inside the O2 plasma cutoff by oscillation analysis of the soft X-ray emission profile using the break-in-slope (BIS) analysis and a harmonic response identification method (HRIM), which is the demonstration of resonant EBH in TCV. The BIS and HRIM methods are also used to successfully detect and track the time-varying deposition locations of one and then two X2 power beams simultaneously. All experimental results are in good agreement within 10% of the normalized plasma radius with numerical results of the AMR and C3PO ray-tracing codes coupled to LUKE. The global power absorption coefficient of modulated ECH (MECH) is studied by HRIM analysis of the plasma toroidal flux response measured by TCV diamagnetic loop (DML). Analysis of earlier X3 MECH and new X2 MECH experiments reveals a major perturbation of the method by the sawtooth magnetohydrodynamic activity in the plasma center. Indeed, an asymmetric improvement of the X3 power absorption (up to 100%) with respect to the sign of the X2 ECCD pre-heating was observed in previous TCV experiments and remained unexplained by Fokker-Planck simulations until now. The present work allows to attribute this asymmetry to a sawtooth activity strongly destabilized by the central X2 co-ECCD locking to the X3 power modulation. The performances of EBW current drive (EBCD) in TCV are studied with the AMR-LUKE codes for several poloidal positions of the EBW injection. The maximum EBCD efficiency is obtained when the EBW are injected close the plasma midplane such that the wave parallel refractive index upshift is moderate and the absorption takes place at the plasma center where the electron temperature is the highest. However, the absolute driven current remains small (i.e. ≲1% of the Ohmic current). Finally, a new loop-antenna for the detection of the lower-hybrid (LH) waves generated by a parametric instability (PI) at the X-B mode conversion was designed, built and installed in TCV torus. Fast monitoring of the LHPI spectrum allows to show for the first time the correlation of the amplitude of the detected waves with the local LHPI threshold power at the mode conversion region, estimated from the experimental profiles data. In an EBW power scan, the LHPI threshold power is estimated to be ≲50 kW in good agreement with the value predicted from the experimental profiles data. The LHPI energy cascade from the low to the high LH frequency bands with increasing EBH power is shown for the first time in TCV.