Ce diaporama a bien été signalé.
Nous utilisons votre profil LinkedIn et vos données d’activité pour vous proposer des publicités personnalisées et pertinentes. Vous pouvez changer vos préférences de publicités à tout moment.

Lecture 8: IoT System Models and Applications

239 vues

Publié le

EEEM048/COM3023- Internet of Things

Publié dans : Formation
  • Soyez le premier à commenter

  • Soyez le premier à aimer ceci

Lecture 8: IoT System Models and Applications

  1. 1. 1 Lecture 8: IoT System Models and Applications EEEM048/COM3023- Internet of Things Prof. Payam Barnaghi Centre for Vision, Speech and Signal Processing (CVSSP) Electrical and Electronic Engineering Department University of Surrey Autumn 2018
  2. 2. Spatial Data and IoT − Most of the data in IoT applications is location dependent; − Services could be also location-based; − Location can be specified as: − Names, labels − Tags and semantic annotations − GPS data - Longitude, Latitude (and Altitude) − How can we define an area? − Multiple points are required − Simple Euclidian distance measure won’t work for longitude/latitude (lon/lat) data − The same way, you can’t simply cluster, group lon/lat data using classical methods that use Euclidian distance (e.g. k-means clustering)
  3. 3. How to create location tags? − GeoHashing is one way to do this; − Geohash is a latitude/longitude geo-coding that was invented by Gustavo Niemeyer. − GeoHashing function can encods/decods (lat,lon) pairs in a compact form. − The Geohash algorithm can represent geographic regions in a hierarchical structure. − A geohash is represented as a string: − e.g. (-25.382708 and -49.265506) can be represented as: 6gkzwgjzn820 − Or http://geohash.org/6gkzwgjzn820
  4. 4. How does it work? − A Geohash is calculated by interleaving bits obtained from latitude and longitude pairs and converting the bits to a string using a Base-32 character map. − Base-32: Table source: https://en.wikipedia.org/wiki/Geohash
  5. 5. Geohash – example 5 ezs42 01101 11111 11000 00100 00010
  6. 6. GeoHash − A Geohash string represents a fixed spatial bounding box. − For example, the latitude and longitude coordinates of (- 25.382708 and -49.265506) fall within the Geohash bounding box of "6gkzwgjzn820". − Appending characters to the string would make it refer to more precise geographical subsets of the original string. − More information: http://geohash.org/site/tips.html#format
  7. 7. Example of similar locations − (51.236127 -0.574036) - Guildford − gcpe6zmbpfrd − (51.243113 -0.590343) – University of Surrey − gcped8d0u087 − (51.243603 -0.587994) – BA (Arthur C Clark) Building − gcped8egdezy gcpe6zmbpfrd gcped8d0u087 gcped8egdezy
  8. 8. GeoHash Prefix similarity can be used to find close locations; But it cannot be directly converted to a metric distance measure.
  9. 9. GeoHashing – Location Codes Image credit: Pramod Anantharam et al., Wright State University/University of Surrey; Alternatively Grid boxes and tags can be defined manually or using other different techniques; here is an example:
  10. 10. Limitations of GeoHash − Geohash algorithm can be used to find locations (e.g. points) that are close to each other by checking prefix similarity of the GeoHash tags. − However, the points that close to each other but located at two opposite sides of the Equator line or for the nodes that fall on line of longitude (i.e. Meridian points) can produce Geohash codes that have no common prefix. − Point close to North and South Poles can have very different geohashes (in Norht and South Poles close areas can have different latitudes) − Geohash also defines a Bounding Box; this can then result having locations that are close but have different GeoHash codes. − For better proximity searches, the surrounding eight geohashes of a geohash should be calculated but this can make the proximity searches more complicated.
  11. 11. How to calculate the distance between two geohashes or locations − If you have two geohashes, the first step is to decode them into latitude and longitude values.   − Longitude varies between [-180 , +180] − Latitude varies between [-90 , +90] − The Geohash is based on the splitting the interval in 2 (for each of the Longitude and Latitude) at each step and take 0 for the right part and 1 for the left part. 11
  12. 12. Calculate distance between two Latitude/Longitude points − There are a variety of calculations for lati­tude/longi­tude points, with different formulae. − The formulae to calculate distance between two Lat/Lon points are often on the basis of a spherical earth (ignoring ellipsoidal effects) – which is accurate enough for most of the applications. 12 Source: http://www.movable-type.co.uk/scripts/latlong.html
  13. 13. Haversine’ formula − haversine’ formula is used to calculate the great-circle distance between two points  a = sin²( /2) +Δφ cos 1φ ⋅ cos 2φ ⋅ sin²( /2)Δλ c = 2 atan2( √a, √(1 a) )⋅ − d = R c⋅ where φ is latitude, λ is longitude, R is earth’s radius (mean radius = 6,371km); 1 =φ lat1.toRadians(); 2 =φ lat2.toRadians(); = (Δφ lat2-lat1).toRadians(); = (Δλ lon2-lon1).toRadians(); 13Source: http://www.movable-type.co.uk/scripts/latlong.html
  14. 14. Haversine’ formula - Note that angles need to be in radians. - In this module, you don’t need to remember these formulae. The main formulae and also the required conversion formula (e.g. degree to radian) will be given in the exam. 14
  15. 15. Haversine’ formula- sample code 15 Source: http://www.movable-type.co.uk/scripts/latlong.html
  16. 16. IoT Applications −The IoT and M2M? −IoT is a more generic term; M2M focuses on devices and machine-to-machine communications; −Sometimes the terms are used interchangeably; however, M2M is mainly meant for automated interactions between devices and IoT is an umbrella term for describing technologies that allow real world data collection, communication, processing and interactions (machines, devices and human users).
  17. 17. Some of the IoT Application areas 17 − Industrial automation − Smart homes − eHealth − Automotive (navigation, traffic control, vehicle safety, fleet management, etc.) − Smart cities (city automation, intelligent parking, intelligent transport systems, air quality and pollution monitoring, etc.) − Environmental monitoring − Smart grid and smart metering − …
  18. 18. Types of applications − Event detection − Nodes report events and occurrences − Anomaly and outlier detection − Collaboration of nearby and/or remote sensors to detect more complicated events − Pattern detection and anomaly detections in patterns − Periodic monitoring and measurements, information extraction and interactons − Measuring and monitoring and reporting the data − Monitoring and measurement can be triggered by an event − Processing the collected information and implementing (automated) interactions. 18
  19. 19. Types of applications − Approximation and edge detection − Detecting how a physical value (e.g. temperature) changes one place to another; − This can be used to approximate spatial characteristics and map it to an area − For example, in a forest fire, this can be used to approximate the border of actual fire; − This can be generalised to finding “edges” in different boundaries such as space and time. − Tracking − An event source can be mobile; − sensors can be used to monitor and track an object; − Speed and direction of the object can be also estimated. 19
  20. 20. Types of applications − Control and feedback − Using actuators to interact with the environment; − Make a change and the sense and obtain feedback from the physical environment. 20 Physical Environment/ Things Actuator Sensor Controller actuation sensing feedbackcommand
  21. 21. Characteristic requirements − Types of services − Interfaces and interaction models − Autonomy of services − Data processing and information extraction requirements − Service network requirements − Quality of service − Delay and latency − Quality of information − Accuracy and quality measures of the functions (e.g. reliability and accuracy of event detection). 21
  22. 22. Characteristic requirements − Fault tolerance − Reliability and dependability − What happens if a node runs out of power or gets damaged or losses coverage − Redundant deployment − Lifetime − Especially if the nodes rely on limited power − Sometimes it is a trade-off between energy efficiency against the quality of service − Can be defined as the time that first node fails or runs out of energy; or when x% of the nodes fail; or it can be defined as the time that the observed “thing” is no longer covered. 22
  23. 23. Characteristic requirements − Scalability − How efficiently large number of nodes/Things can be supported. − How efficiently the system can respond to large number of events, requests, traffic, etc. − Density − Number of nodes per unit area − Programmability − Planning for an application to see if support for change and dynamic updates required. − Maintainability − Ability to adapt to the changes or to change operational parameters − Or, in some applications ability to access and maintain or replace the nodes or to re-configure them (remotely or locally) 23
  24. 24. Required mechanisms − Multi-hop wireless communications − Transmission range can be short and in some cases multi-hop communication is required. − Energy efficient operations − To save energy and/or increase the lifetime of the network/services. − Auto-configuration − Ability to configure (at least some of) the functional parameters automatically. − Collaboration and in-network processing − Several node collaborate − Parts of the process is performed on the node and/or in the network. − Data-centric solutions − Conventional networks often focus on sending data between two specific nodes each equipped with an address. − Here what is important is data and the observations and measurements not the node that provides it. − Security, Trust and Privacy 24
  25. 25. What is special about IoT applications? − Different applications and various requirements − Interaction and deployment in uncontrolled or less controlled environments − Heterogeneity and scale − Energy and resource constraints − Autonomous mechanisms that are often required; e.g. self-configurability − Security and Privacy issues − Data quality and data processing and analysis requirements − Actuation, feedback and control loop to interact with physical objects/environment over distributed networks. − Mobility 25
  26. 26. 26 Crowdsourced solutions
  27. 27. 27 Smart Cities − Cities: − Cities account for 75% of green house emissions, while only occupying 2% of world surface. − It is expected that the amount of people living in urban areas will double until 2050. − By 2015, 1.2 billion cars will be on the road–making 1 car per 6 person. − Challenge: − More space required − Management of resources and infra structure waste, transportation, … − Climate change − Competitiveness − Crisis management Adapted from: Smart Cities and Internet of Things, Oliver Haubensak ETH-MTEC, ETHZ, May 2011.
  28. 28. Designing IoT applications for smart cities 28
  29. 29. 29 Source LA Times, http://documents.latimes.com/la-2013/ Future cities: a view from 1998
  30. 30. Importance of designing for real problems and challenges 30Source: http://robertluisrabello.com/denial/traffic-in-la/#gallery[default]/0/ Source: wikipedia
  31. 31. 31
  32. 32. IoT applications in smart cities − Traffic management − Waste management − City transport − Noise, air-quality control and monitoring − Emergency services − Security and safety − Infrastructure management − Elderly-care and social care − Smart metering 32
  33. 33. IoT environments are usually dynamic and (near-) real-time 33 Off-line Data analytics Data analytics in dynamic environments Image sources: ABC Australia and 2dolphins.com
  34. 34. Traffic control and sensing − There are various sensing technologies that can be used including: video, sonar, radar, inductive loops, bacons, etc. − Some of these will need cable and some need to be installed on lampposts, etc.. − Some others like inductive loops can be embedded into the transportation road infrastructure. − One application is congestion control by counting (and estimating) the number of passing vehicles and their speed. − Environmental factors can affect the sensors; for example, video cameras are not very helpful during fog and heavy snow or rain. 34
  35. 35. #1: Design for large-scale and provide tools and APIs. #2: Think of who will use the data and how, when you design your models. #3: Provide means to update and change the data models and processing methods. 35 Smart Cities
  36. 36. #4: Design for different audience (data consumers, developers, providers) and think about real impact and sustainability of your solution. #5: Specify (and encourage others to do the same) data governance and privacy procedures, explain the ownership and re-use rules, and give control to the owners of the data. 36 Smart cities
  37. 37. 37
  38. 38. Users in control or losing control? 38 Image source: Julian Walker, Flicker
  39. 39. Smart City: IoT Application Examples
  40. 40. 40 Event Visualisation
  41. 41. A sample smart city application 41
  42. 42. Data Quality Explorer 42 Source: The CityPulse project by the University of Applied Sciences Osnabrück Video demo: https://www.youtube.com/watch?v=Yc1jiB5zdfE
  43. 43. Precision Agriculture − A large farm can exhibit wide spatial diversity in soil types, nutrient content, water/moisture, and other important factors. − Sensors for monitoring temperature, humidity, acidity of soil. − Monitoring for fungal disease; e.g. by monitoring humidity and temperature conditions in the field as well as the wetness of potato leaves the researchers can identify the risk of fungal disease. 43 Source: Dargie and Poellabauer, Fundamentials of Wirless Sensor Networks: Theory and Practice, Wiley, 2010
  44. 44. Healthcare: IoT Application use-case scenarios
  45. 45. IoT for healthcare: Dementia use-case  16,801 people with dementia in Surrey – set to rise to 19,000 by 2020 (estimated) - nationally 850,000 - estimated 1m by 2025 (Alzheimer’s Society);  Estimated to cost £26bn p/a in the UK (Alzheimer’s Society): health and social care (NHS and private) + unpaid care;  IoT devices can provide data that can further analysed to detect agitation, sleep, weight loss, and wandering – all have a big impact on quality of life and wellbeing;
  46. 46. Technical Challenges − Infrastructure − Interoperability, integration − Security − Data governance − Scalability
  47. 47. Device/Data interoperability 47
  48. 48. FIHR4TIHM 48
  49. 49. Gateway Gateway Data Analytics Cloud systems External systems Possible links to other networks Gateway Data-driven and patient centered Healthcare Applications
  50. 50. How does TIHM work? Signals Machine Analysis Human Decision Action 24/7
  51. 51. Personalised thresholds 51
  52. 52. Probabilistic machine learning models 52
  53. 53. TIHM Integrated View (iView) Integrated view monitored by blend of health professionals Working closely with Alzheimer’s Society, NHS, social services and police Freeing up clinicians to focus on treatment
  54. 54. 54 Application requirements in IoT − Healthcare − Service reliability, security, privacy, trust, mobility, lower power consumption, lower delays; − Automotive − Mobility, real-time interactions, quality of services, location tracking; − Smart cities − Reliability, fault tolerance, delay tolerance;
  55. 55. Exercise : A use-case study 55Image source: Mahmoud Meribout, A Wireless Sensor Network-Based Infrastructure for Real-Time and Online Pipeline Inspection, IEEE SENSORS JOURNAL, VOL. 11, NO. 11, NOVEMBER 2011 This diagram shows a leak detection System in a pipeline. Work in groups and identify: -What parameters can be measured to detect a leak? -What type of sensors can be used? -What components can be added to this diagram? -What are the key issues that should be considered in the design? -What type of in-network processes can be done?
  56. 56. Further reading − If you are interested in more information about spatial data, you may want to refer to (this won’t be part of the exam): − Spatial Data on the Web Best Practices: − https://www.w3.org/TR/sdw-bp/ 56
  57. 57. 57 Questions?