A Survey on Big Data Analytics in Mobile Cellular Network

The usage of Mobile traffic network is being rapidly increased nowadays. Different techniques are used for boosting the traffic management and improving the performance in mobile networks. Some techniques are used for managing the network such as Apache Hadoop, Mapreduce, Wireless network virtualization and Information – centric networking. This paper is based on the survey of big data analytics in mobile cellular networks. Factors include likely to various methods for minimizing network traffic and rapid growth in the performance network.

Introduction

Big data is the bulk collection datasets so it is big and complex therefore it becomes difficult to process with the help of traditional data processing methods or application.

Big data is the main high phrase in IT domain and new technologies of personal communication is being increased day by day. Initial requirement of big data started from large companies like Facebook, Google, and YouTube etc. For the purpose of analysis of big amount data which is in structured or even in unstructured format. Therefore big is needed everywhere which is process complex and bulk datasets. With recent development in wireless technologies is increasing in mobile networks and mobile application have become both generators and carriers of huge data. In the ancient times big data is use for structured data that is data is well organised in relational databases and spread sheet. Big data analytics has capability to collect scattered data, for understanding the user’s uses pattern from multiple sectors.

It include the user living habits and the timetable can be inferred from the usage of traffic cover different time periods of day, surfing patterns and frequently visited places or the range of activities can be find out from home location register (HLR) databases. Infrastructure is the significant feature of big data analytics. Real-time infrastructure monitor can be possible through big data analytics. And they can take autonomous and dynamic decision. Service providers process large amount of data which is generated by users i.e. Calls Records, Data Records, SMS etc. on daily basis. Big data helps to analyse these data and can be solve most common problems that are faces my service providers. The rapid increased in the data and mobile networks traffic are handled by hadoop framework and Mapreduce programming model can be proposed and provide the security for high traffic data. Analysing and minimizing such huge traffic hadoop framework is being used widely everywhere.

Literature Survey

Gathering data from various sources is one of the parts of big data. When big data efficiently and effectively processed, and analysed, companies are profound to get better customer service, and improved product and services etc. In 5G mobile wireless networks two techniques are defined in the software that are Wireless network visualization and information centric network (ICN). End-to-end network performance can be improved through ICN techniques with integrating wireless network visualization Visualizing is the concept that allows abstraction of physical computing resources into logical units.

Physical resources in cellular networks consist of spectrum resources and infrastructure resources which include radio access networks (RANs), core networks (CNs) and transport networks. Virtualization is an crucial utilization of wireless sensor networks. Mobile user traffic is one the sensing area which takes the advantage of virtualizing the sensors. There is one way through internet infrastructure can be evolve that is Information-centric networking but it can go away from host-centric paradigm based perpetual connectivity and end-to-end principle.

The crucial needs are accessing named resources – not hosts, scalable distribution through replication and catching, good control resolution/routing and access. Network has ICN native catching function in such manner which allow them to node can cache the contents passing through it for a while and deliver them to requesting users. In the network caching mechanism, there is already replicated and probability of this delivery of this content to the user is increased.

Spectrum is the most important factor in mobile communication and network radio. With spectrum sharing, owned spectrum license’s part or all part can be utilized by multiple operators that is based on the agreement. For instance, operator A and operator B have contract in such way that they have to share spectrum band with each other so that they have more flexible frequency scheduling and diversity gain and that’s resulted in to improved spectrum efficiency and network capacity. This paper shows the study about monitoring of traffic network and analysis of large scale cellular network with Hadoop.

Network traffic monitoring and analysis is for optimizing network resource and improving user’s experience. We present here a large scale network based on Hadoop, open source computing platform for distributed storage and distributed processing on commodity hardware. Hadoop consist of many attractive features such as distributed parallel computing, low-cost scale capability and high fault tolerance.

Important Hadoop based tools are developed by Google Such as mapreduce and pig. Mapreduce is such software framework which is used for parallel processing if bulk amount of data on large clusters. Pig is made of two components, first is language itself that is called Pig Latin and second is runtime environment where Pig Latin programs are executed. The system will be bring up through Hadoop framework.

Efficiently processing 4.2 bytes of traffic data from 123 gb/sec links with high performance and low cost every day. J. Liu, et.al introduced scalable wireless bigdata traffic management and developing a bigdata aware wireless network. Scalable wireless bigdata traffic management include two hybrid network structure and Hybrid signal processing models.

In Hybrid network structure, wireless system can adaptively choose only local processing at base station level or only central processing at control unit level or parallel at processing at both levels that is based on the physical channel conditions and correlations at data content. Hybrid signal processing model has commercial optic link operate at link rate 10Gbps for digital communication over single optical carrier. Therefore; under the front haul link capacity constraints system performance is optimized.

Here for improving the throughput and reliability at both mobile terminals and base station of high speed wireless services MIMO antenna technology extensively used. In this paper novel Jaccard measurement based method has been proposed to recognize cellular device models from network traffic data. This is implemented as exactly same as scalable mapreduce program is implemented and it achieves high accuracy. To identify a particular keyword from unformatted textual HTTP headers to represent a cellular device model, Jaccard-based learning method is well suited for it. It consists of three - step.

The first step is extract all keywords from data which describes a cellular device model. The second step is select small set of candidate keywords by calculating the conditional probability of cellular device to reduce the computational workload. At the last step, a Jaccard coefficient value of each candidate keyword is calculated and keywords with the higher values are selected to represent device model. D. kreutz et.al presents software defined networking (SDN) and its architecture.

Physical separation of network control is defined as software defined networking from forwarding plane and where control plane control several devices. The architecture of SDN is vigorously programmable centrally and agile. SDN has successfully managed to cover the way towards a next generation networking. SDN, scalability and performance of devices provides security and dependability.

Conclusion

This survey has used for analysing the problem in the mobile operator networks. For controlling the traffic management, Hadoop, wireless network virtualization, mapreduce tools are used. And this tools are also helps to improve the performance in the cellular networks. This paper helps in recognizing different techniques used in the mobile cellular networks.

Acknowledgment

I am very grateful to our project head manager Mr Haresh Raikwad for his remarks, suggestions and for providing all the vital facilities like use of internet for personal use during office ours and important books that had been suggested and given to us which were essential. I am also thankful to all team members of department.