A REVIEW OF APPLICATION OF DATA SCIENCE TO FORECAST HARDWARE REQUIREMENT BASED ON DATA GROWTH IN TELECOM NETWORK

Abstract

Telecommunication industry is an evolving industry, with a goal of connecting the world by enabling billions of people across the world being able to communicate in split of a second. Last couple of decades have marked the evolution in communication which has significantly affected the way we socially interact with each other. The spike in reliability and availability of telecom industry has surmounted any other commodity therefore telecom sector has significantly seen huge growth in investment and sheer number of users and devices. Operators in telecommunication are experiencing continuous decrease in profits as during the era of 2G the revenue was primarily dependent on voice traffic and it allowed data traffic in 300+ kbps. Minimalistic cell phone of 2G in combination of small data rates does not offer much data base services. Voice call traffic take up small portion from the network. For example, 2Mbps can hold up to 30 simultaneous voice calls, therefore the network expansions were not very regular. Thus, the CAPEX was less whereas profit margin was soaring as network deployed 5-6 years ago could handle all the traffic. The evolution of 3G and 4G allowed higher data rates thus making it possible for various companies to introduce new product that will replace voice call e.g. Whatsapp, Viber, Line etc. These applications also removed the restrictions of international dialling thus shifting 90-95% of IDD to these apps. Table1 exhibit the impact on bandwidth usage comparison between normal call vs app based call. This shift decreased the revenues and normal calls became more profitable as compared to app-based calls. Activity Bandwidth Throughput/hour MB/hour Voice multiple Normal Voice Call 16.0 56 7 1 Low Data Usage Call (App) 200.0 703 88 1 General Data Usage Call (App) 350.0 1230 154 1 Table 1 Bandwidth Use of Voice and Data Activities Review of Data Science Model for Estimating Data Growth in Telecom Network Increase in data rates on mobile devices have influenced many companies to provide various platforms that rely on data rates to streaming and social media services. Appendix-A Figure 1 exhibit the increase in average time spent on social media since the evolution of 3G and 4G market. These on one hand require high data rates to provide better services but has a drawback on network resources as cost of incurring expansion decreases the overall profit margin of operators while fulfilling the network requirements. Video streaming and social media platforms like Netflix, Youtube, Facebook, TikTok etc. have taken the world by storm people right now check their social media profile prior to even brushing their teeth in the morning. Mobile devices have helped in exponentially increasing the number of content creators on these platforms in past 5 years. Also, mobile devices have been the prime reason behind the consumption of video-based content. Network expansion has the solution for most the problem which arise with ever growing demands to get mobile data service at higher data rates from 3G, 4G and now to 5G. But the network expansion is very difficult to estimate as the basics of economics suggests that there are limited resources therefore, they should utilize to maximize the profitability while providing best service quality to customers. Even though most telecommunication operators are mostly profitable but companies around the world are observing saturation in their revenues thus resulting in declining the overall bottom-line. During 2G and 3G era most of the customer requirements were fulfilled via rollout the network expansion in transport part of the network. Network expansion and rollout are two activities that are performed in repeated cycles to fulfil the customer demands. Rollout involves deployment of hardware on some location for the first time to enhance the coverage and footprint Review of Data Science Model for Estimating Data Growth in Telecom Network of the network whereas network expansion is done on existing sites for capacity building to ensure that network has enough capacity to cater all active users simultaneously. Transport part of the network is not normally shown in the network topologies of Mobile architecture, but they are very crucial in connecting the radio sites with the core devices. Their purpose in telecom network is like roads and highways, without them no one can transverse to the desired location. Planning a transport network properly means that it has enough capacity handle the traffic even in case of some disaster, fiber cuts, or traffic in access to estimated traffic. Equipment ordering cycle takes around half a year from PO issuance till the implementation of the equipment. Therefore, estimation and forecasting become quintessential as in case mistake in calculation the network may have to face congestion most portion of the year. Network congestion does not mean that complete network is facing problem, it can be site-level, cluster level, node-level, region-level. Recent introduction of cloud-based network elements have increased the dependency on the availability of the transmission/transport medium, in currently only a few core sites are left in the network. Following figure 1 shows the Venn diagram can give the relationship of number of Radio, Transport, IP Core, Core sites in telecommunication network. Data science and analytics has emerged as a vital resource to enhance the productivity and profitability. In telecom BI (Business Intelligence) has been embedded in the organization for quite a long time therefore they understand the impact of big data mining and data science as it has helped them to provide offers based on BI inputs to retain and enhance the customers. But unfortunately, organization are not using it to make decisions related to other departments to enhance their estimation and forecasting as most the individuals in other departments lack the skillset required to build synergy between all departments based on data inputs. Review of Data Science Model for Estimating Data Growth in Telecom Network This project is an example which show that effectiveness of data science is not limited