The first direct observation of gravitational waves on the 14th September 2015 opened up a new window to the Universe. Since then, gravitational wave astronomy has provided highly valuable information about previously mostly unexplored astronomic events like the merger of two black holes; nevertheless, it is still in its infancy, with numerous phenomena to be discovered and investigated. This requires improved detectors, featuring higher sensitivities.Seismic noise is among the most relevant noise sources for current gravitationalwave detectors. Although featuring sophisticated seismic isolation systems, currentdetectors are directly and indirectly limited by seismic noise below 30 Hz; therefore, current and future detectors require novel isolation systems and strategies to achieve their design sensitivity.The Albert Einstein Institute (AEI) 10m prototype is a test facility for gravitationalwave detectors to develop and study novel technology. The primary goal is to reach and surpass the interferometric Standard Quantum Limit (SQL) based on quantum noise. It requires significant suppression of all classical noise contributions in order to achieve the design sensitivity. AEI Seismic Attenuation Systems (AEI-SASs) isolate the sub-SQL interferometer against seismic noise and are used to develop and demonstrate novel techniques for gravitational wave detectors. The AEI-SASs combine passive isolation based on the principle of a harmonic oscillator and active isolation based on feedback loop suppression.In the scope of this thesis, the active seismic isolation of the AEI-SASs is implemented and analyzed in detail. Two different isolation strategies are described,namely local seismic isolation and global seismic isolation. The former is predominantly used by current gravitational wave detectors and focuses on inertial isolation of each interferometer component individually. The latter is a mostly untested principle implemented in the AEI 10m prototype in a unique realization. It focuses on the minimization of differential motion between the interferometer components.All involved sensors are characterized, and their noise is calculated. The sensornoise is measured in so-called huddle tests with excellent agreement to the models, providing detailed insight into limitations of the sensors and the entire system. Requirements for precise huddle tests are investigated in measurements.The application of local isolation techniques at the AEI 10m prototype is describedand novel methods for their improvement are demonstrated. An enhanced coordinate system transformation increases the decoupling between different degrees of freedom, resulting in better isolation performance. The sensitivity dependence on the sensor alignment is analyzed in a novel approach, enabling the capability of a sensitivity optimization and an improved system characterization. The combination of different sensors to exploit their most sensitive frequency regimes is optimized using a new calculation method. It is adapted for the purpose of global isolation by including inter-platform sensors and the coupling of motion of the globally isolated AEI-SASs to the sub-SQL interferometer. This method simultaneously provides detailed information about limitations of the isolation system, which is used to propose and analyze possible improvements for the global isolation. An optimization of digital filters is calculated to improve the seismic isolation by a factor of 1.4. A realistic upgrade of vertical sensors offers an additional improvement of up to a factor of 4.3. Some fundamental statements of global isolation are confirmed by measurements.Furthermore, a noise budget of the AEI sub-SQL interferometer is simulated. Requirements on noise suppression and interferometric parameters are set by comparison to the SQL. Based on these requirements, a possible design for the anti-symmetric port photodetector is motivated, and its noise is analyzed.