Knowledge extraction from sensitive data often needs collaborative work. Statistical databases are generated from such data and shared among various stakeholders. In this context, the ownership protection of shared data becomes important. Wa-termarking is emerging to be a very effective tool for imposing ownership rights on various digital data formats. Watermarking of such datasets may bring distortions in the data. Consequently, the extracted knowledge may be inaccurate. These distortions are controlled by the usability constraints, which in turn limit the available bandwidth for watermarking. Large bandwidth ensures robustness; however, it may degrade the quality of the data. Such a situation can be resolved by optimizing the available bandwidth subject to the usability constraints. Optimization techniques, particularly bioinspired techniques, have become a preferred choice for solving such issues during the past few years. In this paper, we investigate the usability of various optimization schemes for identifying the maximum available bandwidth to achieve two objectives: (1) preserving the knowledge stored in the data; (2) maximizing the available bandwidth subject to the usability constraints to achieve maximum robustness. The first objective is achieved with a usability constraint model, which ensures that the knowledge is not compromised as a result of watermark em-bedding. The second objective is achieved by finding the maximum bandwidth subject to the usability constraints specified in the first objective. The performance of optimization schemes is evaluated using different metrics.