Contents I. Introduction
The Internet is growing at a very fast rate and reaching to every corner of the globe. As per the report of Statista, the global digital population as of January 2021 is 4.66 Billions [1]. With such popularity of the internet, new internet related technologies are evolving such as Internet of Things and Cyber Physical Systems. Moreover, smart devices such as Google Home, Amazon Echo and Smart Mesh WiFi Systems are growing in demand in the digital market worldwide. Therefore these days any network consists of traditional end devices such as PC, laptops, mobile phones and tablets as well as smart devices such as Amazon Echo and Google Home. This results in the generation of humongous and heterogeneous network traffic. Analysis of such network traffic is done by network administrators, security analysts, network architects, network researchers and security researchers for constructive technology development. In order to analyze network traffic, network traces are first captured in pcap format using sniffing tools like tcpdump [2] or tshark [3] and then analyzed. However analysis of such network traces is hard because heterogeneity and humongousness of current network traffic. The tools that are available in the market for network traffic analysis impose three major challenges. First challenge is that a wide variety of powerful tools available in the market are paid tools such as Solar winds [4] and Auvik [5]. Second challenge is that free tools such as PacketTotal [6] and A-Packets [7] usually comes with a limited set of functionality. Third challenge is that available tools mostly extract packet level features only [8]–[10] and misses flow level features. Moreover, flow based feature extraction is computationally expensive process because of packet concatenation which belongs to the same flow based on packet-id, source IP address, destination IP address, source port number, destination port number and layer 4 protocol.
