The discovery of a population of high-redshift dust-obscured submillimeter galaxies (SMGs) from ground-based submillimetre (submm) cameras has revolutionised our understanding of galaxy evolution and star formation in extreme conditions. They are the strongest starbursts in the Universe approaching the Eddington limit, and are believed to be the progenitors of the most massive galaxies today. However, theoretical models of galaxy evolution have even been unable to explain the large number of detections of high-redshift SMGs. A very few among them are gravitationally lensed by a foreground galaxy. Recent wide-area extragalactic surveys have discovered hundreds of such strongly lensed SMGs, opening new exciting opportunities for observing the interstellar medium in these exceptional objects. We have thus carefully selected a sample of strongly gravitational lensed SMGs based on the submm flux limit from the Herschel-ATLAS sample. Using the IRAM (Institut de Radioastronomie Millim$\acutee$trique) telescopes, we have built a rich H$_2$O-line-detected sample of 16 SMGs. We found a close-to-linear tight correlation between the H$_2$O line and total infrared luminosity. This indicates the importance of far-IR pumping to the excitation of the H$_2$O lines. Using a far-IR pumping model, we have derived the physical properties of the H$_2$O gas and the dust. We showed that H$_2$O lines trace a warm dense gas that may be closely related to the active star formation. Along with the H$_2$O lines, several H$_2O^+$ lines have also been detected in three of our SMGs. We also find a tight correlation between the luminosity of the lines of H$_2$O and H$_2O^+$ from the local ULIRGs (UltraLuminous Infrared Galaxies) to the high-redshift SMGs. The flux ratio between H$_2O^+$ and H$_2$O suggests that cosmic rays from strong star forming activities are possibly driving the related oxygen chemistry. Another important common molecular gas tracer is the CO line. We have observed multiple transitions of the CO lines in each of our SMGs with the IRAM 30 m telescope. By analysing the CO line profile, we discovered a significant differential lensing effect that might cause underestimation of the linewidth by a factor of sim2. Using the LVG (Large Velocity Gradient) modelling and fitting the multi-J CO fluxes via a Bayesian approach, we derived gas densities and temperature, and the CO column density per unit velocity gradient. We then found a correlation between the gas thermal pressure and the star formation efficiency. We have also studied the global properties of the molecular gas and its relationship with star formation. We have derived the gas to dust mass ratio and the gas depletion time of our sample galaxies, which show no difference compared with other SMGs. With the detections of atomic carbon lines in our SMGs, we extended the local linear correlation between the CO and CI line luminosity. Finally, we compared the linewidths of the CO and H$_2$O emission lines, which agree very well with each other. This suggests that the emitting regions of these two molecules are likely to be co-spatially located. In order to understand the properties of molecular emission in high-redshift SMGs, and more generally, the structure and the dynamical properties of these galaxies, it is crucial to acquire high-resolution images. We thus observed two of our brightest sources with the ALMA (Atacama Large Millimeter/sub-millimeter Array) and NOEMA (NOrthern Extended Millimeter Array) interferometers using their high spatial resolution configuration. These images have allowed us to reconstruct the intrinsic morphology of the sources. We compared the CO, H$_2$O, and dust emission. The cold dust emission has a smaller size compared with the CO and H$_2$O gas, while the latter two are similar in size. By fitting the dynamical model to the CO data of the sources, we have shown that the sources can be modelled with a rotating disk, and the projected dynamical mass and the effective radius of those sources have been obtained. With the future NOEMA and ALMA, we will be able to extend such kind of observations to a larger sample lensed SMGs and even to unlensed SMGs, to study various gas tracers, and to refresh our understanding of the physical conditions of the ISM and their relation to the star formation.