Research Projects & Publications

As the team leader at Gridsentry Private Limited, I designed an advanced Intrusion Detection System (IDS) to safeguard the IEC 61850 based digital substations. This advanced IDS not only scrutinizes multicasted GOOSE messages for conventional signatures of tampering but intricately validates them against the unique electrical fingerprints of substations. This innovative approach ensures not only the safety of communication streams but also fortifies the integrity of critical information, redefining the boundaries of substation security.

Phasor Measurement Units (PMUs) are the most advanced sensing devices deployed in the smart grid to monitor grid operations continuously. PMUs had been assumed secured against cyber-attacks; however, recent research shows that data integrity attacks can also be launched against the PMUs, exploiting the basic governing laws of electricity flow, to influence the AC state estimation process. Hence in this project, an affine relationship between the sparsely placed PMUs has been extracted using the support vector regression (SVR), which was not possible through the deterministic approaches. This affine relationship has been used to identify and flag the PMUs with false data in their measurement stream. Further, SVR models have been used to determine the amount of corruption in flagged PMUs, to correct the measurements at the control center.

The smart grid's vulnerability against possible cyber attacks can be improved by replacing the least secure measurement devices (\textit{e.g.} RTUs) with more secured sensors (\textit{e.g.} PMUs). Thus, a dynamic programming based approach has been proposed in this work to determine the multi-stage sensor deployment plan. This multi-stage deployment will distribute the capital cost requirement over a time-horizon while alleviating power system vulnerability against cyber-attacks with each deployment phase. The Attack Feasibility Index (AFI) is used to determine the sequence of sensor placement in the grid, within a deployment stage, based on each substation's exposure to a possible cyber attack. The proposed scheme also ensures the grid's critical observability in its first stage while improving the observability level with each subsequent phase.

A defender needs to know the extent of the CPS exposure against such attacks before deploying the smart grid's countermeasures to fight against cyber-attacks. An Index is proposed in this work to quantify the probability of data integrity attacks on a power system. This index is then used to compare the relative vulnerability of different smart grid components, thus assessing the smart grid's vulnerability against cyber attacks.

An attack vector is a matrix that expresses the amount of coordinated corruption introduced in each field measurement to obtain the desired outcome via a data integrity attack on the AC state estimator. Previously available methods were iterative and hence was time-consuming. Therefore, a new approach was proposed in this project, which linearised the attack vector formulation process so that a credible data integrity attack can be launched against the nonlinear state estimator in minimal time.

Voltage and current phasors are the complex quantities measured by the PMUs and sent to the control centers to monitor the smart grid's secured operation. It has been reported that sometimes measurements from some of the PMUs did not reach the control centers for various reasons, making the operator unable to take critical decisions. Thus, an ANN-based approach has been proposed to discover the hidden pattern between the PMU measurements and predict the missing measurements at the control center. The concept was initially tested on the IEEE 39-bus test system's synthetic measurement data and finally verified on India's norther regional power grid's real-world data.

PMUs are advanced sensing devices that can measure the voltage phase angles along with voltage magnitudes, which was impossible before, thus improving the grid monitoring by multifold. However, because of their high cost, they can not be placed on each of the nodes of the electric grid. Hence, in this work, an Integer Linear Programming (ILP) based optimization approach has been proposed, which will minimize the number of PMUs, while maximizing the smart grid observability by placing the PMUs on strategic nodes.