Observability invention - A critical challenge in AI Data Center expansion

With the rise of artificial intelligence, the network infrastructure connecting different parts of a data centre is as crucial as the processing power of the data centre. The underlying engineering problem is that operators need to observe, diagnose and validate high-capacity optical systems without disrupting the services these systems provide.

Praveen Kumar Pal is a researcher and an inventor in the field of advanced optical communications and has solved this problem with an invention called “Observability Without Changing the Target.” The whole idea behind the technology is one of strong principles – gaining visibility into a system without having to change the system under observation. That distinction is very real for data centre interconnect (DCI) networks where reliability and uninterrupted operation are paramount.

This is the international platform for networking research using AI. It is an international platform for AI-enabled networking research.

Mr Pal was chosen as an Opening Keynote Speaker for the International Conference on Intelligent, Secure & Human-Centred Technologies (ICISHCT 2026) to be held at the Institute of Advanced Research, Gandhinagar, India on 29th and 30th October. The invited keynote speakers were Dr Samar Muti from Liwa University, UAE, and Dr Dhaval Pujara, Provost, Adani University, India. His nomination brings to the table the perspective of industry inventors and academic leaders at an international forum.

His professional record also encompasses research publications, technical peer review and optical communications, distributed systems and intelligent network management. These activities are part of ongoing practice and research on the engineering issues related to the next-generation network infrastructure.

Mr Pal has addressed the critical and chronic industry problem in his Keynotes-

Why observability matters for AI-scale data centres

AI infrastructure does not scale through computing capacity alone. Data centres must exchange enormous volumes of information across high-speed optical links, and operators must bring new connections online while maintaining the availability of existing ones. As networks expand across sites, vendors and generations of equipment, troubleshooting becomes more difficult: engineers need dependable insight into performance and faults, often under strict change-control and uptime requirements.

Traditional troubleshooting can require intrusive diagnostic steps, software changes or maintenance coordination. In a production environment, each intervention can add engineering effort, delay problem isolation and create operational risk. An observability approach that avoids changing the target addresses that bottleneck directly. It allows monitoring and validation to be considered separately from disruptive changes to the equipment or applications under observation.

The significance extends beyond an individual device. When an operator can examine network behaviour with less disruption, teams are better positioned to diagnose faults, verify upgrades and accelerate readiness checks across complex infrastructure. These capabilities are especially valuable in DCI environments using high-capacity coherent optical technologies, including 400ZR and ZR+, where network availability and multivendor operations must coexist.

An invention directed at a practical industry constraint

Mr Pal’s Patent on “Observability Without Changing the Target” USPTO: US11943121B2 reflects a design philosophy shaped by the demands of mission-critical communications: network visibility should not depend on first altering the environment that engineers need to inspect. By addressing that constraint, the invention offers a technically relevant path toward more efficient fault isolation, operational validation and management of sophisticated optical systems.

This is important for the pace of AI data centre construction. Building a facility is only part of bringing new AI capacity into service; the facility must also be connected, tested, monitored and integrated into a dependable network. Every avoidable diagnostic intervention can slow that process. Nonintrusive observability can help reduce such friction, supporting more repeatable network turn-up and more confident expansion of interconnected facilities.

Mr Pal’s work therefore speaks to a broader industry challenge: how to scale optical infrastructure quickly without sacrificing the reliability expected of critical digital services. The invention’s technical relevance lies in helping reconcile those competing requirements, rather than treating rapid expansion and dependable operations as separate goals.

From technical invention to wider industry relevance

The industry significance of nonintrusive observability is clear: as AI increases pressure to deploy high-capacity data centre networks rapidly, operators need ways to understand network behaviour without introducing additional operational risk. Mr Pal’s invention addresses that underlying engineering need and highlights the role of software and systems architecture in making large-scale optical infrastructure manageable.

The next phase of AI growth will depend not only on faster optical transmission, but also on how reliably those networks can be deployed, validated and maintained. In that context, Mr Pal’s focus on observability without changing the target represents an important technical direction for the evolution of data centre interconnect operations.