Developments in information and communication technology have made it realistic to produce data at high rate, resulting to stress communication and computational infrastructure and making it difficult to store and transmit that data. To deal with these unbounded and continuous data streams is a sheer challenge for researchers from communication, data storage, computational and data mining domains because of its continuity, unbounded in nature, endless arrival and concept drifting with the passage of time. Extracting useful and hidden information from these ubiquitous data streams is one of the major goals during the last decade. Both supervised and unsupervised techniques of knowledge discovery were being researched e.g. clustering and frequent pattern mining. Frequent patterns reveal important and hidden information in the data and there are many application areas where these can be very helpful to improve overall performance of the system. Aims: Main objectives of this research are three-fold. Firstly our target is to review the existing scientific and analytical techniques addressing data mining in ubiquitous and continuous streams of data. Secondly, addressing data collected from ubiquitous devices and investigating device resources for local or centralized mining in streaming data environment. Finally, our focus is to devise comparatively efficient and accurate methodology for finding closed frequent itemsets in streaming data. Similarly, incorporation of contextual information in mining process is also addressed. Methods: As a first step of this research, we performed a regress review of existing scientific techniques and algorithms specifically designed to extract hidden information from the ubiquitous data streams and studied limitations and problem areas to develop a new and improved version of existing techniques. Then we performed analytical study on ubiquitous data mining resource restriction and limitations and comparison of centralized and distributed data mining in ubiquitous environment. Regarding the final aims of this research, we have developed a generic framework that is adaptable, scalable and incorporating contextual information to improve data mining results and outcomes.