Bad Data Injection in Smart Grid Attack and

Bad Data Injection in Smart Grid: Attack and Defense Mechanisms Zhu Han University of Houston

Overview l l Bad Data Injection l Power System State Estimation Model l Introduction to Smart Grid Defender Mechanism – Quickest Detection l Attacker Learning Scheme – Independent Component Analysis l Future Work l A Few Topics in Smart Grid Communication l Conclusions l Quick View of Amigo Lab

“Smarter” Power Grid l l Sensing, measurement, and control devices with two-way communications between the suppliers and customers. Benefits both utilities, consumers & environment: – – Reduce supply while fitting demand Save money, optimal usage. Improve reliability and efficiency of grid Integration of green energy, reduction of CO 2 l More than 3. 4 billion from US federal stimulus bill is targeted. – Obama stimulus plan l One of hottest topic in research community – But what are the problems from signal processing, communication and networking points of view?

Are more easily integrated into power sys. Less depend on fossil fuel Gather, monitor the usage so the supply more efficiently and anticipate challenging peaks Realtime analysis, Manage, plan, and Smart Grid forecast the energy in-home meets the system to management tool needs usage to track Use sophisticated comm. Technology to find/fix problems faster, enhancing reliability Connect grid to charge overnight when demand is low Can generate own and sellback excess energy

Supervisory Control and Data Acquisition Center l Real-time data acquisition – Noisy analog measurements u Voltage, current, power flow – Digital measurements l State estimation – Maintain system in normal state – Fault detection – Power flow optimization – Supply vs. demand SCADA TX data from/to Remote Terminal Units (RTUs), the substations in the grid

Privacy & Security Concern l More connections, more technology are linked to the obsolete infrastructure. – Add-on network technology: sensors and controls estimation – More substations are automated/unmanned l Vulnerable to manipulate by third party – Purposely blackout – Financial gain – Story of Enron How to tackle this issue at this moment? Provide one example next

Power System State Estimation Model l Transmitted active power from bus i to bus j – High reactance over resistance ratio – Linear approximation for small variance – State vector , measure noise e with covariance Ʃe – Actual power flow measurement for m active power-flow branches – Define the Jacobian matrix – We have the linear approximation – H is known to the power system but not known to the attackers

Bad Data Injection and Detection l l State estimation from z Bad data detection – Residual vector – Without attacker where – Bad data detection (with threshold ) without attacker: with attacker: otherwise l Stealth (unobservable) attack: z=Hx+c+e, where c=H x – Hypothesis test would fail in detecting the attacker, since the control center believes that the true state is x + x.

Overview l l Bad Data Injection l Power System State Estimation Model l Introduction to Smart Grid Defender Mechanism – Quickest Detection l Attacker Learning Scheme – Independent Component Analysis l Future Work l A Few Topics in Smart Grid Communication l Conclusions l Quick View of Amigo Lab

Basics of Quickest Detection (QD) l Detect distribution changes of a sequence of observations l as quick as possible l with the constraint of false alarm or detection probability. min [processing time] l s. t. Prob(true ≠ estimated) < ŋ Classification 1. Bayesian framework: l known prior information on probability l SPRT (e. g. quality control, drug test, ) 2. Non-Bayesian framework: l unknown distribution and no prior l CUSUM (e. g. spectrum sensing, abnormal detection )

QD System Model l Assuming Bayesian framework with non-stealthy attack – the state variables are random with l The binary hypothesis test: l The distribution of measurement z under binary hyp: (differ only in mean) l We want a detector – False alarm and detection probabilities

Detection Model - Non. Bayesian l Non-Bayesian approach – unknown prior probability, attacker statistic model l The unknown parameter exists – in the post-change distribution and may changes over the detection process. – You do not know how attacker attacks. l Minimizing the worst-case effect via detection delay: l We want to detect the intruder as soon as possible while maintaining PD. Detection delay Detection time Actual time of active attack

Multi-thread CUSUM Algorithm l CUSUM Statistic: How about the unknown? where Likelihood ratio term of m measurements: l By recursion, CUSUM Statistic St at time t: l Average run length (ARL) for declaring attack with threshold h Declare the attacker is existing! Otherwise, continuous to the process.

Linear Solver for the Unknown l Rao test – asymptotically equivalent model of GLRT: l The linear unknown solver for m measurements: l Recursive CUSUM Statistic w/ linear unknown parameter solve: – Modified CUSUM statistics The unknown is no long involved

Simulation: Adaptive CUSUM algorithm l 2 different detection tests: FAR: 1% and 0. 1% l Active attack starts at time 5 l Detection of attack at time 7 and 8, for different FARs

Markov Chain based Analytical Model l Divide statistic space into discrete states between 0 and threshold – Obtain the transition probabilities – Obtain expectation of detection delay, false alarm rate and missing probability

Overview l l Bad Data Injection l Power System State Estimation Model l Introduction to Smart Grid Defender Mechanism – Quickest Detection l Attacker Learning Scheme – Independent Component Analysis l Future Work l A Few Topics in Smart Grid Communication l Conclusions l Quick View of Amigo Lab

Independent Component Analysis (ICA) l Linear Independent Component Analysis – find a linear representation of the data so that components are as statistically independent as possible. – i. e. , among the data, find how many independent sources. l Question for bad data injection: – Without knowing H, the attacker can be caught. – Could attacker launch stealthy attack to the system even without knowledge about H? – Using ICA, attacker could estimate H and consequently, lunch an undetectable attack.

ICA Basics l A special case of blind source separation u=Gv l u = [ui, i = 1, 2, … m]: observable vector l G = [gij, i = 1, 2, … m, j = 1, 2, … n]: mixing matrix (unknown) l v = [vi, i = 1, 2, … n]: source vector (unknown) l Linear ICA implementation: Fast. ICA from [Hyvärinen]

Stealth False Data Injection with ICA l Supposing that the noise is small, then we what to do the mapping: u=Gv z=Hx l Problem: state vector x is highly correlated l Consider: x = A y, where – A: constant matrix that can be estimated – y: independent random vectors l Then we can apply Linear ICA on z = HA y – We cannot know H, but we can know HA – Stealthy attack: Z=Hx+HA y+e

Numerical Simulation Setting l Simulation setup – 4 -Bus test system, IEEE 14 -Bus and 30 -bus – Matpower

Numerical Results l MSE of ICA inference (z-Gy) vs. the number of observations (14 -bus case).

Performance of the Attack The PDF is the same w or w/o attacking. So log likelihood is equal to 1– unable to detect

Overview l l Bad Data Injection l Power System State Estimation Model l Introduction to Smart Grid Defender Mechanism – Quickest Detection l Attacker Learning Scheme – Independent Component Analysis l Future Work l A Few Topics in Smart Grid Communication l Conclusions l Quick View of Amigo Lab

1. Distributed Smart Grid State Estimation l l The deregulation has led to the creation of many regional transmission organizations within a large interconnected power system. A distributed estimation and control is need. – Distributed observability analysis – Bad data detection l Challenges: – Bottleneck and reliability problems with one coordination center. – Need for wide area monitoring and control – Convergence and optimality

Fully-Distributed State Estimation l With N substations/nodes Local observati on matrix Local Jacobian matrix Useful information to be detected Unknown State – By iteratively exchanging information with neighbors – All local control center can achieve an unbiased consensus of system-wide state estimation.

2. Optimality of Fault Detection Algorithm l Detecting the attack as an intermediate step towards obtaining a reliable estimate about the injected false data – Facilitates eliminating the disruptive effects of the false data l Joint estimation and detection problem – Define an estimation performance measure – Seek to the optimize it while ensuring satisfactory of the detection performance Performance measurement

3. Manipulate Electricity Market Example: Ex Post Market that recalculate optimal points for generation and consumption based on real-time data Min : St: Generation Cost Power Balance Generation & Transmission limits [28]

4. PMU l PMU can measure voltage angle directly – Defender: placement problem, no need to place nearby – Attackers’ new strategy with existence of PMU 1 PMU 2 PMU 3 PMU 4 5 PMU 6 7 PMU [29]

5. Game Theory Analysis (attacker, defender) N A N (0, 0) D (c, -c) a, b, c (b, -b) (-a, a) How to formulate the game? t

A Few Topics in Smart Grid Communications l Bad data injection l Demand side management – Peak to average ratio – Scheduling problem l Renewable energy – The renewable energy is unreliable. – Have to use diesel generators during shortage – Not cheap and not green l PHEV – routing, scheduling and resource allocation l Communication link effect on the smart grid

Conclusions l l Bad data injection problem formulation From defender point of view – detect malicious bad data injection attack as quick as possible – Adaptive CUSUM algorithm l From attacker point of view – can estimate both the system topology and power states just by observing the power flow measurements – Independent component analysis algorithm to obtain information – Once the information is at hand, malicious attacks can be launched without triggering the detection system l l l Many possible future work Edited book 2012 by Cambridge with E. Hossain and V. Poor. Possible future collaboration

Overview of Wireless Amigo Lab l Lab Overview – 7 Ph. D. students, 2 joint postdocs (with Rice and Princeton) – supported by 5 NSF, 1 Do. D, and 1 Qatar grants Current Concentration – Game theoretical approach for wireless networking – Compressive sensing and its application – Smartgrid communication – Bayesian nonparametric learning l – – – l Security: trust management, belief network, gossip based Kalman Physical layer security Quickest detection Cognitive radio routing/security Sniffing: femto cell and cloud computing USRP 2 Implementation Testbed

Questions Thank you for listening and supporting!