179 CHAPTER 7 SUMMARY AND CONCLUSION This chapter summarizes our research achievements and conclude this thesis with discussions and interesting avenues for future exploration. The thesis describes a novel architecture for data and transaction management in distributed mobile cloud systems based on Surrogate Objects. To realize mobile cloud paradigm for data and transaction management, various issues such as connectivity, heterogeneities of resources, latency issues, load fluctuations on mobile and static devices, data availability and failures in devices and network connectivity were handled in different perspective. Thus in the process of exploring the new architectural design for data and transaction management in distributed mobile cloud, the integration of mobile devices into the cloud and cloud federation were also investigated and experimented. An effective Data and Transaction management system is essential in this emerging mobile cloud paradigm which requires effective communication primitives by considering the constrained nature of the cloud, and mobile devices participated in the paradigm. To realize such a data and transaction management in mobile cloud, a novel architecture called Surrogate Object based Mobile Transaction (SOMT) was proposed and experimented initially at the distributed mobile system level. The proposed data management model was further explored for data mining process for an effective knowledge discovery mechanism, namely, Surrogate Object based Data Mining (SODM). Thus the integration
180 of mobile devices into data mining process using Surrogate Object was investigated and experimented. Finally, the integration of SOMT and SODM was explored to provide a novel solution for the constrained mobile devices participating in the mobile cloud, providing an efficient technique for data and transaction management, namely, Integrated Obj_FedRep (IOFR) for cloud federation and replication. Contributions to these three paradigms have been outlined in detail. Performance studies for all the three paradigms were conducted to test the effectiveness of the proposed data and transaction management system in distributed mobile cloud systems. The key issues related to each of the paradigms were identified and appropriate model were designed to provide a solution to most of the identified issues. The SOMT was investigated to deal with the fundamental issues of asymmetry problems, latency problems, low abort rate and disconnection in distributed mobile transaction executions. The proposed strategy caches the data in the surrogate object which helps in reducing the average transaction time. The required reliability was provided by sending the transaction request to surrogate object and network lifetime was maximized by migrating the surrogate objects in a way to avoid improper load balancing. The model also provides an elegant solution to disconnected transaction and higher cost under heavy load conditions in mobile environment by buffering client requests or using the cached data to handle them. The surrogate object can remain active, maintains information regarding the current state and play an active role on behalf of the device and reduces the network congestions, overcomes the asymmetry in wired and wireless access and achieves the low abort rate. This has made data and transaction execution on distributed mobile systems more reliable and feasible. The model has been verified and validated by implementing surrogate object technique on the top of existing kangaroo transaction model over a simulated distributed mobile paradigm. The simulation prototype also compares the performance of the proposed SOMT
181 model with an existing other mobile transaction models. The performance showed that the number of data access, wireless and wired access, abort rates over the distributed mobile network in the proposed surrogate object model was very less compared to all the existing models. The demand for anytime, anywhere connectivity for an emerging data mining applications has emerged to support context based ubiquitous mining. In recent years, there is a tremendous growth in the development of mobile devices and wireless technologies; however, there are still major limitations, especially to support both the disconnection and mobility. Mobility additionally brings the major challenge of unreliable, low, and variable bandwidth connectivity. The existing techniques for data mining process may not be adequate because it requires a tight cooperation between mobile devices and mining server. The second paradigm of this thesis assessed all these difficulties involved in the ongoing data mining application in distributed mobile systems, and developed an effective knowledge discovery mechanism; namely, Surrogate Object based Data Mining (SODM). The protocol defines a method that allows mobile devices to participate seamlessly in mining process and act on behalf of mobile device and permit to cache the location based frequent datasets and retrieve the knowledge from those datasets. The protocol reduces the number of wireless and wired data transfers, overhead imposed on the mining server and solves totally the location management problems. The performance of the proposed SODM technique has been studied and compared with and without presence of surrogate objects. The result of the simulation show that, the constraints of the distributed mobile system for data mining applications were reduced and an effective mining capacity for an information processing was increased. The distributed mobile systems mentioned in the previous two paradigms for data and transaction processing can be taken to the next level
182 by providing seamless wireless extensions to the mobile cloud. Mobile device can share resources through federation and replication modes and overcome the limitations and provide data, computing as utility services to the underlying transaction applications in mobile cloud. The integration of SOMT and SODM was proposed to provide a novel solution for the constrained mobile devices participating in the mobile cloud, providing an efficient technique for data and transaction management, namely, Integrated Obj_FedRep (IOFR) for cloud federation and replication. This proposed research model federates two or more service providers for the purpose of load balancing and cloud replication for fast and easy access. It creates a realtime ongoing clone of database server of different service providers using surrogate object in the mobile support station and an exact copy updated with the database server in synchronized and unsynchronized manner. The model handles the federation and replica at the object level which in turn provides better response time and minimizes the overall network traffic incurred due to mobility and database server failover. The novel feature of the proposed model was handling the constraints of mobile devices and issues in federation and reducing the communication overhead in both wired and wireless network. Thus the proposed model tolerates failures and limitations of the mobile devices and provides a significant reduction in wireless access, abort probability and response time. The performance of IOFR has been evaluated and compared with an existing federation protocols by simulation. The results of the simulation show that proposed object based mobile cloud model enables to solve computationally intensive task for data and transaction processing by pooling the cloud resources to surrogate objects located in multiple clouds through federation techniques. This model provides reliable integration of SOMT and SODM to mobile cloud platform through object federation and replication techniques and provides elegant support for various constraints of
183 the mobile devices and cloud environment such as poor computational resources, limited energy power, low bandwidth, unpredictable consumer request, limited services facility offered by the cloud, overloaded consumer request during peak time, and problems due to mobility under various conditions. Thus, the results of the simulation show that IOFR has lesser latency, lesser abort rate, reduced wireless and wired access, reduced bandwidth utilization and is more scalable and fault tolerant. 7.1 FUTURE RESEARCH DIRECTIONS There is still work needed to be carried out in our IOFR data and transaction processing system in Mobile Cloud.. Thus, given the huge size of the computing facilities backing the service offering of cloud providers, energy efficient solutions play a fundamental role in mobile cloud. Such an enhanced solutions will be useful for a large scale transaction processing systems. One of the possible scenarios where we can create smarter and greener data processing design, involves an efficient placement of virtual machines over the surrogate objects, and develops energy driven reconciliation server management. Moreover, IOFR support the implementation of data and transaction processing system by deploying the infrastructure and defining surrogate object model enabling the interoperability among multiple clouds maintained by multiple service providers and federation heavily rely on multiple clouds from different cloud vendors by themselves. Therefore, with the advent of cloud federation, we can invent a new model in which service consumers will be able to leverage multiple clouds. This implies the service consumers directly compose and make use of other clouds services and add more trust to access the data in mobile cloud paradigm.