Recently, I came across an online article that said that the planet has 4.3 billion mobile phone users1 . One reader even made an interesting comment about how the high number of mobile signals would help the extra-terrestrials discover us faster! The bit about the spike in the number of signals is of course true as almost every person I see: kid, teen or adult, wields a mobile phone today. Voice and data usage volumes have shot up thanks to the higher usage, better coverage and a packed portfolio, adding significantly to big data volumes. I think this and the fact that the space is turning highly competitive are bringing about rapid changes in a related space: telecom analytics. Demand for real-time analytics especially is huge and the spotlight today is on big data in motion.
Along with machine-level embedded intelligence built within the networking equipment, I see other dimensions of analytics insights manifesting in various ways. It is heartening to see analytics being used for diverse purposes ranging from fraud detection to enrichment of user experience. With the declining ARPU for traditional voice services and flat rate data charges, Communication Service Providers (CSPs) are looking at avenues for new revenue earning applications. I have noticed that CSPs in developed markets have begun to monetize their information troves by offering new
lifestyle services in collaboration with other utility providers like retail stores, event management companies, and security agencies. However, I think that there are a few pressing challenges that need to be addressed in order to strike a balance between the privacy of users and exploitation of data to derive meaningful insights. Lack of awareness of implications on implementing a solution may expose a CSP to privacy-related laws and regulations. For instance, storage of data that may reveal user identity in a data store that is not strictly controlled would render the solution vulnerable to potential attacks. I therefore believe that data anonymity should assume high priority.
While anonymity is good from security perspective, gaining an insight into end-to-end services often necessitates a common key for cross correlation of events in an analytics engine. That could be a significant challenge when there are multiple data sources of different vendor equipment. As an example, let us assume that I’m using my mobile phone in the course of my work commute. As I move from one locality in my city to another, data correlation should take place in tandem with the changing connectivity topology, which unfortunately, does not always happen. For accurate analytics insights, there is a need to find ways to correlate data drawn from different individual silos even as the context changes. Over dependence on the vendor for network troubleshooting and high processing time in deep packet inspections are some of the other issues I’ve come across. Content filtering on particular network nodes, say preventing adult content downloads on a specific node, can be computationally intensive. Besides, I find many CSPs struggling to perform runtime analytics at the network service level.
Challenges apart, I believe that there are opportunities for innovations in the telecom ecosystem that will help exploit big data and analytics to its full potential. CSPs can develop ‘Self-organizing Self-optimizing Networks’ (SON) that use real-time insights without human intervention. For instance, if network resources are underutilized, some resource configuration could be shrunk dynamically to save power. I also think that there are opportunities to address any customer experience related issue in real time with emergence of real time analytics at the network service level. Emerging technologies like Software-Defined Networking (SDN) will open up the network and this will simplify decoding of complex network events. As the decoding becomes simple, running analytics across multiple hops of the network to derive end to end insight will be greatly simplified.