<

Fahad Javed
Department of Computing
School of Electrical Engineering and Computer Science (SEECS)
National University of Science and Technology (NUST)

CRIMSON Lab, H-12
Islamabad
Pakistan
fahadedupk@gmail.com
fahad.javed@seecs.edu.pk

Research Areas


  • Hyper-personalized learning
    • Hyper-personalized learning is defined as data driven andragogy for every learner. Andragogy is the learner centric learning which pivots the experiences and expertise of the learner to enhance her knowledge, understanding and skill to achieve a mutually acceptable goal of the learner and the trainer. This is usually targeted to adult learning but with availability of MOOCs and re-evaluation of the traditional educational paradigms, HPL has the potential to be the next big thing in education. In our design [See fig] we see this as a hierarchical control loop with primary loop focusing on measuring student’s engagement with the current content. This loop feedbacks to the learner’s personalized model which adapts to the style and abilities of the learner. The top level loop controls the learning path, of what one wants or needs to achieve and the systems responses with the subject, topic and abilities to achieve the overall goal. We are testing the proof of concepts of the first control loop to measure student engagement.
    • Measuring Student Engagement
      • A plethora of technologies are being proposed in the literature for the future consumers. However, what will be acceptable or preferable to the consumer has not been investigated in depth. Furthermore, there are technologies which the consumer is willing to use but evalaution of its impact on the supply-demand equation is difficult to gauge. In this project we are building a what-if simulator which simulates the changes in electric grids demand based on the consumers acceptance and response to new devices, technologies and policies. The simulator divides the population into different socio-economic societal groups. Individual societal groups are then modeled independently where devices within a house act as agents to simulate the consumers behavior with new technology. This project is closely associated with the willingness to invest and use survey we conducted for prosumer modeling
      • Research Team:
        1. Fahad Javed (Lead)
        2. Alishba Tahir [Pscych eval and sensor fusion]
        3. Ahsan Ijaz [Platform and LMS]
        4. Syed Iftikhar Naqvi [Platform and LMS]
        5. Salsabeel [Data collection app]
        6. Hina Fareed [Cloud infrastructure]
        7. Mhwish [Self-awareness in LMS system]
        8. Ali Sarwar [Gesture identification]
        7. Usman [Speech processing]

    • Prosumer modeling: characterizing consumers, Forecasting loads, eliciting elasticity
      • The future energy scenario is envisioned as a complex system where the consumer of today will be much different. The high penetration of smart devices and renewable and distributed energy sources means that the consumer of today will be an active player in the grid with ability to purchase, buy and sell electricity on the grid in real-time. To do all this, it is important to know what the prosumer better. The research in this stream is dedicated to modeling and characterization of the consumer of today to make him a more robust and efficient prosumer of tomorrow.
      • Willingness to Invest/Use Next-gen Energy Technologies
        • One of the most important questions in the domain of energy is that what how will the consumers respond to new energy technology. In this study we survey urban, peri-urban and rural areas in Pakistan to identify the willingness of the consumers to 1. participate in the various state of the art energy management strategies and 2. invest in various renewable and smart energy technologies. From this study we will build a what-if simulator to support the decision makers in energy sector by providing simulations on how the consumers will respond to certain policy changes.
        • Research Team:
          1. Fahad Javed (Lead)
          2. Dr. Tasneem Akhtar
          3. Maria Zaffar
          4. Usman Ali
      • Consumption Characterization
        • In this study we are trying to prove that energy consumption of a household is an anthropologic-structuro-temporal phenomenon. Our basic hypothesis is that the consumption at different times depends on different variables of the system. That is, it may be the case that in mornings the number of household occupants define the load but in the middle of night the type walls impact the consumption with minimal impact of occupant count. We apply various data-mining, clustering and statistical analysis techniques to prove, or disprove, this hypothesis. The output of this research has serious consequences for household level forecasting of energy needed for planning DSM. The research follows from the result of STMLF forecasting for household levels as STMLF implicitly points out to such anthropologic-structuro-temporal nature of household load.
        • Research Team:
          1. Fahad Javed (Lead)
          2. Arisha Akas
          3. Adeela
          4. Ahmad Taimoor
      • Forecasting
        • The electric grid is changing. With the smart grid the demand response (DR) programs will hopefully make the grid more resilient and cost efficient. However, a scheme where consumers can directly participate in demand management requires new efforts for forecasting the electric loads of individual consumers. In this research we try to find answers to two main questions for forecasting loads for individual consumers: First, can current short term load forecasting (STLF) models work efficiently for forecasting individual households? Second, do the anthropologic and structural variables enhance the forecasting accuracy of individual consumer loads? Our analysis show that a single multi-dimensional model forecasting for all houses using anthropologic and structural data variables is more efficient than a forecast based on traditional global measures.
          Our next step in this research is to see the generalizability of our initial results. The goal of this task is to see if we can generate a model general enough for it to represent loads from houses which are not used for generating the model.
        • Research Team:
          1. Fahad Javed (Lead)
          2. Khurram Usman
          3. Muhammad Hassan Jaffry
      • Elasticity elicitation
        • Elasticity is the range of operation of a device which is acceptable for the consumer. The core idea of demand side management revolves around this elasticity, or range within which we can move the device consumption. In this study we explore ways to identify this range by observing the historical usage of the consumers.
        • Research Team:
          1. Fahad Javed (Lead)
          2. Taha Hassan
          3. Muhammad Zubair
          4. Ali Hasaan

    • Demand Side Management
      • Demand side management is the process of impacting the demand of the electricity for better management of resources. We aim to resolve three goals through DSM. First is to mitigate supply-demand gap by surgical, fair and optimal load shedding. Second is optimizing the usage of renewable energy sources (RES) in residential areas by moving the loads to maximize utilization. The third is leveling peak demand by moving the elastic loads forward or back to reduce the need for generating units in a given horizon.
      • CDRS: Energy cost-reduction solution for the commercial consumers
        • Demand shaping in residential and small-scale commercial establishments presents one significant way to better manage growing power demand and reduce greenhouse gas emissions. However given the large number of end-users with relatively low individual energy demands and a variety in appliance usage preferences, demand shaping for residential and commercial consumers presents a challenge. To deal with such concerns and fluctuations in daily power demand, we propose a cooperative demand response system which is supplemented with energy storage. We build a demand response algorithm using basic probability and control theory concepts to achieve robust and smooth demand regulation. Our algorithm leverages available energy storage in a power system to reduce peak demand without negatively impacting end-user's life-style. To illustrate the validity of our algorithm, we setup a regulated power network in an office and conduct empirical experiments. We also collect power demand profiles for selected appliances and test the algorithm in computer simulation. Results validate that the system achieves demand regulation within a reliable error margin, scales to large power networks, accommodates end-user preferences, and ensures fairness in local decisions.
      • Managing high consumption devices in city/region
        • Shortage of electricity is a major problem in many developing countries. Unfortunately, for some of these countries the only solution to this problem is to shut down complete electricity supply to a few neighborhoods to make up for the gap between demand and supply. To this end, we have developed a self-optimization approach to reduce the gap between demand and supply through remotely controlling high powered electric devices such as air conditioners. In this approach we have used mathematical optimization techniques such as linear programming to intelligently manage the electricity distribution. Not only through this approach we have been able to provide service-level guarantees to the consumers but we have also shown that our approach is fast, scalable and has the ability to handle unscheduled spikes in the system.
      • Holistic energy management in the micro-grids
        • Economic dispatch and demand side management (DSM) are two of the most important tools for efficient energy management in the grid. It is a casual observation that both these processes are intertwined and thus complement each other. Strategies aiming to optimize economic dispatch have implications for demand side management techniques and vice-versa. These tasks - although intrinsically linked - have traditionally been con- ducted independently due to the size of the grid and lack of surgical controls and information flow. However, in micro-grids, where the spread of power is relatively small and control over demand and supply is more robust and fine grained, it becomes possible to tie these two tasks together and arrive at a more globally optimal solution for energy management. In this work we present a genetic algorithm-based solution which combines economic dispatch and demand side management for residential loads in a micro-grid. Our system collect preferences of demand data from consumers and costs of energy of various sources. It then finds the optimal demand scheduling and energy generation mix for the given time window. Our evaluations show that the given approach can effectively reduce operating costs in a single and multiple-facility microgrids for both suppliers and consumers alike.
        • Research Team:
          1. Ahmer Arif
          2. Fahad Javed (Lead)
          3. Naveed Arshad
      • CDRSR: Integrating renewable with CDRS
        • In this work we integrate renewable sources of energy (RES) with the CDRS algorithm. The CDRS works on the principle of moving the peak loads to level the maximum consumption level for better economic dispatch. But the case of RES requires maximization of consumption at times of RES availability. The CDRSR is an extension of CDRS to incorporate this change
        • Research Team:
          1. Fahad Javed (Lead)
          2. Raheela

    • Simulation and Data management
      • Deploying and validating demand side management requires some ancillary and supporting tools. In this research stream we look at the support tools for DSM. Specifically we focus on simulation and data management.
      • Simulating Consumer Response to Policy and Technology Intervention
        • A plethora of technologies are being proposed in the literature for the future consumers. However, what will be acceptable or preferable to the consumer has not been investigated in depth. Furthermore, there are technologies which the consumer is willing to use but evalaution of its impact on the supply-demand equation is difficult to gauge. In this project we are building a what-if simulator which simulates the changes in electric grids demand based on the consumers acceptance and response to new devices, technologies and policies. The simulator divides the population into different socio-economic societal groups. Individual societal groups are then modeled independently where devices within a house act as agents to simulate the consumers behavior with new technology. This project is closely associated with the willingness to invest and use survey we conducted for prosumer modeling
        • Research Team:
          1. Fahad Javed (Lead)
          2. Adnan Shafiq
          3. Aqeel Hanif
          4. Maria Zafar
      • Customizable Granular City-level Device Simulation
        • Smart grids provide newer ways of energy production, transmission, and distribution. In a smart grid finer control of electrical devices in household and buildings are implemented to better manage energy demand and supply. However, this finer control commonly known as demand side management (DSM) requires extensive simulation at various levels before a DSM algorithm may actually be deployed in a real building or neighborhood. Since forecasting the energy usage behavior of myriad number of electrical devices is a difficult exercise, simulations are done to assess the effectiveness of a DSM algorithm. The problem with the state-of-the-art simulators is that each is designed for simulating electrical devices� behavior under specific and limited settings. To this end, we present a highly configurable and extensible smart grid simulator (SGS) that is capable of simulating per-minute granularity of energy usage under numerous settings. Moreover, SGS is able to simulate behavior at four levels: electrical devices, households and buildings, neighborhoods and cities. Given different scenarios SGS can simulate relativistic behavior of energy usage at all four levels.
        • Research Team:
          1. Fahad Javed (Lead)
          2. Usman Ali
          3. Naveed Arshad
      • Stream-based Device Simulation
        • Efficient energy management is considered a key to human progress. One of the most promising techniques to achieve this in the future smart grid is demand response. To evaluate demand response and prove it efficacy requires a simulation platform capable of constructing different scenarios before its deployment. The goal of such a simulation will be two fold, first to simulate the response of the system against the stimuli and second to produce states which are valid. In this work we use a case based reasoning (CBR) simulation framework. We show that this CBR based simulation model is more accurate than the existing machine learning and regression based models. We show through our results that CBR can accurately construct a model with fewer training example with virtually no analytical complexities.
        • Research Team:
          1. Fahad Javed (Lead)
          2. Maria
      • Data Compression
        • The smart grid is the next generation of electricity system in which information technology will aid in providing more efficient, secure and robust energy management. To facilitate this, automated metering infrastructure is being deployed to capture household energy consumption. The state of the art meters can collect this data at the granularity of a second or even lower. To transmit, store and manage this fast streaming large quantities of data is a challenging problem. Although different algorithms exist for compaction and compression of data, managing time series data of this magnitude and peculiarity is a less explored problem so far. In this work we apply three different data management techniques that can be applied to the smart metering data. We explore different algorithms that can be used and the issues that arise from their usage on the smart meter data. We also look at some peculiarities of the smart metering data which can be harnessed for more efficient data management. In the end we provide an analysis of the techniques and a guideline for selection of algorithms for smart meters data management infrastructure.
        • Research Team:
          1. Fahad Javed (Lead)
          2. Muhammad Nabeel
          3. Zaid bin Tariq
          4. Rana Waqas
          5. Danish Jalil
          6. Naveed Arshad

    • Self-managing Systems
      • Self-management has been the underlying technology in most of my research work. From this research work there are three streams of works which contribute to overall state of the self-managing systems.
      • Adaptable Optimization
        • Optimizing self-evolving and dynamically changing systems is a grand challenge. In order to apply optimizations almost all conventional optimization techniques require a runtime system model. However, system models and their solution techniques vary in their strengths and limitations. For a rigid system, a single system model is acceptable. But if the system is constantly changing its structure then a rigid model is not able to represent the system properly, resulting in an inefficient use of technique in some cases. Therefore, in this work we propose a framework for an optimization engine that adapts the optimization technique based on the system state. The adaptation involves selection of techniques based on historical statistics and current data, and dynamic generation of a model at runtime. This runtime model is then used to apply a relevant optimization technique to find a desired optimization plan for the system. We have evaluated the proposed framework on an electricity distribution system. Our results show that the proposed framework is adaptable, fast and able to manage numerous situations.
      • Self-Calibration
        • Autonomic and autonomous systems exist within a world view of their own. This world view is created from the training data and assumptions that were available at their inception. In most of these systems this world view becomes obsolete over time due to changes in the environment. This brings a level of inaccuracy in the autonomic behavior of the system. When this degradation reaches a certain threshold self-healing or self-optimizing systems generally recreate the world view using current data and assumptions. However, the self-optimization process is akin to kill a fly with a hammer for minor tuning of the world view. Instead we propose the idea of self-calibration for self-managing these systems. We define self-calibration as the ability of the system to perceive the need for and the ability to execute minimal tuning to bridge the gap between system's world view and incoming information about the outside world. In this we present a case for considering self-calibration as a self-* enabling property of systems specifically for time-critical systems using data-centric AI technologies. We present our case by discussing three case studies from different domains where self-calibration enables a system to become self-healing or self-optimizing. We then place self-calibration in a generic system and explicitly describe the types of systems in which self-calibration can be implemented and the benefits that one can accrue from its inclusion.
        • Research Team:
          1. Fahad Javed (Lead)
          2. Malik tahir Hassan
          3. Khurram Junejo
          4. Asim Karim
          5. Naveed Arshad
      • Self-Optimizing Tag Recommender for Social Bookmarking Systems
        • In this work, we propose and evaluate a self-optimization strategy for a clustering-based tag recommendation system. For tag recommendation, we use an efficient discriminative clustering approach. To develop our self-optimization strategy for this tag recommendation approach, we empirically investigate when and how to update the tag recommender with minimum human intervention. We present a nonlinear optimization model whose solution yields the clustering parameters that maximize the recommendation accuracy within an administrator specified time window. Evaluation on "BibSonomy'' data produces promising results. For example, by using our self-optimization strategy a 6% increase in average F1 score is achieved when the administrator allows up to 2% drop in average F1 score in the last one thousand recommendations.
        • Research Team:
          1. Fahad Javed (Lead)
          2. Malik tahir Hassan
          4. Asim Karim
          5. Naveed Arshad


Research Interests

My research focus is on developing practical solutions for the developing world. This has resulted in my exploration of different fields ranging from investigation of anthropological impact on energy consumption to understanding student engagement parameters and attempting to capture them through technology. I group my current research into three broad streams: Educational technology, Prosumer modeling, Demand side management.