Dynamic Watermark Injection in NoSQL Databases

With the advent of gathering real time information, the need of Not Only SQL (NoSQL) databases has skyrocketed. No prior work is done on watermarked protection of such Schema-less databases. Traditional watermarking techniques cannot be applied as NoSQL databases are dynamically increasing and often have irregularities in schema. In this paper, a new perspective of embedding watermark into such databases is proposed. Watermark that acts as a signature is securely prepared for each tuple individually. The proposal deals with the issues by leveraging the flexible schema features provided by NoSQL database. A new attribute is dynamically injected into the tuple before inserting or updating it in database. Experimental results and analyses demonstrate that with even the slightest modification of up to 5% alteration in tuples, the watermark recreated from the suspected database changes by as much as 20%, thus facilitating immediate detection with localization up to tuple level in the database. Keywords: Fragile Watermarking; Tamper Detection; NoSQL databases; Big Data; Dynamic Watermark Injection