Projects

Harnessing WiFi for Soil Moisture Detection in Smart-Agriculture

Smart-agriculture strives to use technology to increase the quality and quantity of agricultural output while guaranteeing sustainability. Key to achieving these objectives are “connected” sensors that monitor a wide variety of farms’ health indicators. Arguably the most important of these indicators is the soil moisture content. Indeed, such measurements must be acquired with a high temporal and spatial resolution so as to realize efficient utilization of irrigation resources. The most commonly utilized system for soil moisture content measurement relies on a network of conventional contact sensors placed inside the soil. However, achieving high spatial resolution with this approach requires the sensors to be spread throughout the farm, thus rendering the approach infeasible for vast fields. An alternate approach is to use ground penetrating radars mounted on mobile platforms. However, these platforms are bulky and expensive, and thus infeasible for resource-constrained farms such as those in Pakistan.

The objective of this initiative is to develop a soil moisture sensing system that is scalable to vast fields, achieves good spatial resolution, and at the same time is cost-effective. Exploiting the fact that the reflection properties of radio frequency (RF) waves are a function of the moisture content of the reflecting surface, we intend to develop a sensing system composed of an off-the-shelf WiFi transceiver transmitting and receiving radio signals in the unlicensed ISM band. Such an RF-based system is all-the-more attractive since the system could make the moisture measurements using the same piece of WiFi hardware that it employs for communicating the measurements to a centralized smart farming controller. In addition, since WiFi transmissions are typically made with devices that are cheap and light, the system, when mounted on a mobile platform, would promise cost-effective soil-moisture measurements with high spatial resolution.

A Mobile Decision Support System for Antenatal Healthcare in Rural Settings

World Bank statistics puts the infant mortality rate (IMR) in Pakistan at a staggering 6.9% in 2013 while the data from World Health Organization for the same year puts the maternal mortality rate (MMR) at 170 per 100,000 live births. Deplorably, these statistics are only partially depictive of the scale of the problem as Pakistan does not have a civil registration process for data on maternal mortality. Consultations with urban and rural clinical practitioners in Pakistan reveals that high mortality rates are only reflective of the prodigious vacuum in access to essential antenatal and maternal health care facilities. In this project, we aim to improve antenatal health-care access in rural Pakistan by developing a mobile antenatal diagnostics platform powered by an intelligent and automated decision support system (DSS) based on the NICE guidelines of the National Health Service, UK. The DSS will be implemented on an android tablet, and will be integrated with wireless health monitoring devices for providing antenatal clinical markers. Upon deployment, we envision the proposed mobile platform to ensure that consultant-quality advice is served to rural patients upon request or via push notifications in the case of high-risk pregnancies. At the same time, the clinical health markers acquired through the wireless sensors will also provide significant assistance in forecasting and addressing antenatal anomalies. The project therefore is set to serve as an initiative towards providing quality antenatal healthcare advice to rural areas of Pakistan.

Development of a Software Defined Radio Test-bed utilizing GPS Signals for Navigation Applications

Among many satellite based navigation systems, American global positioning system (GPS) is unique as it is fully operational providing different kind of services for civil and military usage in different frequency bands. This research project dealt with the design and development of a basic software receiver exploiting only main civilian GPS frequency band also known as L1 C/A GPS signal. The focus of the project was to develop a platform which can be used in the future not only for research activities but also for the development of related positioning and navigation applications. A fully software defined GPS receiver has been developed which not only provides the user position, velocity and timing information but also has the flexibility to provide access to the raw data within the receiver at different points. In this way, this platform allows us to develop and test not only novel algorithms but also develop innovative GPS based positioning applications which otherwise were not possible to develop due to lack of flexibility attached with hardware receivers, generally available in the market. The receiver is fully software based and coded in C language. Its performance has been tested and validated by collecting the real GPS data using front-end in different environments.

Acoustically Green Zones: Design and development of an active control system for noise reduction in both open and closed spaces

Acoustic noise pollution has become one of the major environmental issue which has detrimental effects especially in urban areas. Urban noise sources such as construction work, industrial activity and traffic noise etc. are affecting a large population in the form of producing direct and cumulative adverse health effects such as hearing loss, cardiovascular disease, sleep disturbance, tinnitus, annoyance, working and learning disabilities etc. This project aimed at developing a working simulation system and related theory for acoustic quiet zone generation by cancelling undesirable noise and acoustic interference over a volumetric region in open or closed spaces. Our approach has been limited to developing theoretical framework and then testing this framework using simulations. We have been successful in obtaining some initial results where the noise cancellation has been achieved over a spatial region as shown below.

KneaBEAT: Design and development of a platform to monitor human knee health

Monitoring the health of human knee is of utmost importance as degraded knee health severely impacts the quality of life of individuals by reducing their level of activity for long periods of time and even increase morbidity. With younger population, knee injuries are very common among athletes, military personnel, and other populations engaged in high performance activities. With relatively older population, different joint related disease affect the knee health severely. . In this research work, we aim to develop a prototype solution for effective monitoring of knee health. Sounds produced by the knee likely depend on the angle of the bones, severity of degradation, lubrication and wear of cartilage. Our proposed solution relies on these acoustical emissions from human knee during complex motion patterns. These sounds will be recorded by a contact microphone inside wearable solution along with motion sensors to measure the orientation of knee during each sound event. These recorded signals will then be used by advanced developed signal processing algorithms to extract relevant information from them and classify them according to knee pathology for effective diagnosis and monitoring of knee health.

A Low cost, high accuracy and improved integrity cooperative driver assistance platform for enhancing traffic safety and road network efficiency

Every day, we hear about the road traffic accidents, traffic jams, delays in deliveries, vehicle collisions, pedestrian’s injuries and other issues related to two basic problems regarding road transportation: road-user-safety and road-network-efficiency. These two problems are affecting almost every person living in cities and are becoming worst with every passing day due to increasing urbanization of Pakistan cities. As a result, we are paying not only in terms of human tragedies, societal frustration, and bad quality of living but also in terms of economy. The proposed research aims to address these problem technically by developing a platform to be deployed inside vehicles which are connected with each other over a wireless link. In this cooperative scenario, each vehicle can sense and perceive its environment more intelligently and can alert the drivers about a possible event. This event could be safety critical e.g. chance of a collision with nearby vehicle or could be efficiency critical e.g. traffic jam in next round-about. We are aiming a low cost, highly accurate solution with an extra degree of robustness in the form of integrity and reliability of all the information.

Design and Development of a Wearable Device Based Diagnostic and Rehabilitation Tool for Parkinson’s Disease

About 281 Million people in the world suffer from some kind of tremor related disorders; among them 10 Million suffer from Parkinson’s disease (PD) for which there is no good diagnostic or rehabilitation tool. Since, PD is not curable, an early detection has its particular importance in the sense that the treatments available for controlling different symptoms are most effective at this stage. Also, PD is very difficult to diagnose as there is no single test for it and it requires examinations by doctors who have expertise in movement disorders. Doctors use different subjective tests for its diagnosis and one such test costs $300 and takes almost an hour to complete. These tests usually work well when the patient is in later stages of PD. For a doctor, there is nothing that can help him in decision aid for PD and give him quick results especially in early stages of PD. Because of the lack of a clear diagnostic test, the symptoms are misdiagnosed as other neurological disorders mostly. Once therapy such for PD is started, it leads to development of motor fluctuations such as wearing off, early morning dystonia, delayed-ON response etc. which requires daily monitoring of physical activities for objective and valuable clinical information regarding the type and level of activities of the patients and their changes during medication. PD affects the quality of life of patients severely as most of them fail to perform their daily routine tasks properly and lose control over their will to do anything due to tremor severity. In this project, we aim to develop a decision support system coupled with multiple sensors to help the medical practitioners in diagnosing PD at its very early stage.

Data driven statistical learning of system dynamical models based on GP regression framework

Gaussian processes have been proven to be powerful probabilistic non-parametric modeling tools for static nonlinear functions. However, many real-world applications, such as control, target tracking, and time-series analysis are tackling problems with nonlinear dynamical behavior. The use of GPs in modeling nonlinear dynamical systems is an emerging topic, with many strong contributions during the recent years. In this project, we are exploring the dimensions of GP regression in identification of state space models of a dynamical system using input output data.

Active Prosthesis Device: design and development

The loss of a limb is a major disability. Unfortunately, today’s prosthetic technology is a long way from realizing fully functioning artificial limb replacements. Although lower-extremity prostheses are currently better able to provide assistance than their upper-extremity counterparts, very basic locomotory problems still remain. For example, compared with intact persons, walking amputees require 60% more metabolic energy depending on walking speed, physical fitness level, cause of amputation, amputation level, and prosthetic intervention characteristics. Additionally, amputees walk at 11-40% slower self-selected gait speeds than do persons with intact limbs. Such clinical problems may, in part, be attributed to today’s prosthetic ankle-foot designs. Commercially available prostheses comprise spring structures that store and release elastic energy throughout each walking stance period. Because of their passive nature, such prostheses cannot generate more mechanical energy than is stored during each walking step. In distinction, the human ankle performs positive network and has a greater peak power over the stance period, especially at moderate to fast walking speeds. Active knee prosthesis solves this problem along with other issues too. We are developing an active knee prosthesis and an active hand prosthesis within our group.

Development of Algorithm and Prototype Hardware for Estimation of Crease Geometry using Image Processing and Computer Vision Techniques

In Tetra Pak production process, the creasing is applied to the paperboard to enable good and precise folding of the board. Creasing helps in getting precise, stronger and high-speed folding of the packaging material produced by Tetra Pak in the filling machine at the customer end. The process of creasing is embedded with the printing line, where the creasing is applied through male die pressing the paperboard into a female die after the printing on the paperboard. The creasing process is important in a sense that it produces permanent deformations in the board. These deformations weaken the board, and consequently make it easier to fold, resulting in a package with durable edges and high strength. In this project, we aim to develop algorithm and prototype hardware for the estimation of crease geometry using image processing and advanced signal processing methods. The developed algorithm will be generating depth profile of the crease on the paperboard.

Advance Signal Processing Methods for the Estimation of Ground Water Storage Variations in Indus River Basin using GRACE Data

"With over 60 per cent of Pakistan's water pumped from natural underground reservoirs, and no limits on how many wells can be drilled and how much they can be drained, it becomes difficult for authorities to precisely monitor the underground water levels." This statement signifies the necessity of groundwater monitoring in Pakistan where most of the urban population consume groundwater. Continuous growth of population and cities without planning and/or without groundwater monitoring, industrial growth and increase in agricultural cultivation will result in inevitable water resource and environmental problems (not only for big cities) due to constant decrease of water table. The conventional groundwater monitoring methods are not only time and money consuming, but also limited by their spatial coverage, and therefore cannot produce dynamic observations over the large spatial region. In this project we aim to monitor dynamic changes in groundwater storage in Indus River Basin (IRB) by processing the satellite data. We are using NASA's Gravity Recovery and Climate Experiment (GRACE) satellite mission data as it has been previously used in conjunction with data obtained from Hydrological models to monitor groundwater storage changes in large river basin on a seasonal/annual scale. The GRACE data provides an estimate of terrestrial water storage (TWS) variations. In order to remove soil-water changes and snow water component from TWS to get groundwater storage (GWS) change, we are using data provided by NASA's Global Land Data Assimilation System (GLDAS).

Link to my talk on the project: Click here

Development of Algorithms and Hardware for Innovative Features for Digital Vending Machines

We are carrying out this project in collaboration with Digital Retail Pvt. Ltd. (DR) that is involved in developing, selling and operating Digital Vending machines in Pakistan. In this project, we are designing algorithms and prototype hardware for the development of innovative features for digital vending machines. These innovations may potentially enhance the experience of the users of the vending machines by capturing customer data and using it to make decisions and performing appropriate actions. The development of the innovations outlined in this proposal will serve as an an opportunity for SBASSE to engage its students in real life challenges and provide them a test bed for solving industry problems.

Harnessing Spherical Geometry in Scientific and Engineering Data Processing

Scientific analysis that takes into account the spherical geometry of the phenomenon in question is becoming increasingly important in many application domains, for example:

• The melting of polar ice sheets is a major contributor to global sea-level rise. Estimating the trends of the spatial distribution of (for example) Antarctica’s ice mass loss can be recovered from processing satellite gravity data.

• The distribution of galaxies in the Universe is being provided by new generation of satellite surveys that will map large areas of the sky with unprecedented detail in many wavelengths. The revealed large scale structure informs our understanding of the Universe and how it is evolving.

• The neuronal pathways in the brain reveal the data highways and our thoughts in action with time. They are being mapped with increasing resolution as we refine our imaging technologies based on magnetic resonance imaging (MRI) and diffusion tensor imaging (DTI).

In research into Signal Processing aspects, we focused in this project to

• develop advanced, fast and robust signal processing methods for data collected with spherical geometry (spherical signals) building on our recent published work organized as follows:

o framework for the spatio-spectral analysis and development of computationally efficient algorithms for signal processing in the spherical harmonic and joint spatio-spectral domains;

o optimal and efficient sampling schemes on the sphere and the rotation group with associated fast spherical harmonic transforms;

o efficient signal representations and robust processing methods that respect signal spatial anisotropy, exploit signal sparsity and take into account spin-valued and tensor-valued signals; and

o low complexity and computationally efficient kernel-based representations of spherical signals as an alternative to the higher dimensionality of spherical harmonic based methods.

• apply the developed tools in real world applications, especially to the processing of:

o diffusion tensor imaging (DTI) and magnetic resonance imaging (MRI) data in neuorimaging,

o head related transfer function (HRTF) measurements in acoustics, and

o cosmic microwave background (CMB) data in cosmology.

Green Communications: Toward Developing an Energy-Efficient Platform for Internet of Things (IoT)-enabled Devices

Most of the ubiquitous devices in the IoT paradigm are inherently resource constrained. As a result, the design requirements for these next-generation networks demand that they be capable of reliable wireless communications at a negligible computational energy cost. Thus they need to strike the right balance between communication reliability/throughput and transmission plus computational energy budget. The objective of this project is to develop transmitter receiver architectures that achieve the optimum tradeoffs between these parameters.

Smart Data, Systems and Applications

Contact Details

School of Science & Engineering
Lahore University of Management Sciences

tahir@lums.edu.pk

(+92) 42 35608000 Ext. 8423, 8177, 8112