Please enable JavaScript.

Coggle requires JavaScript to display documents.

Privacy (Attacks to location privacy (Query sampling attack (Assumptions,…

- - - - Active devices such as mobile phones cannot preserve stealth
      - Access of information overrides stealth
  - - - Vulnerability to data mining: a person's identity can be inferred from his or her location over time
      - A barrier to authentication and personalisation
      - Efficacy in sparcely populated areas?
    - - Analysis of microdata
      - Quasi identifiers: sufficiently correlated with other quasi identifiers to form a unique identifier
      - Databases: k-anonymous if every record in the table is indistinguiable from at least k-1 other records with respect to quasi identifiers
    - - Spatial cloaking: adapt the spatial precision of location information about a person according to the number of other people in the same quadrant
      - k-anonymity: individuals are k-anonymous if their location information cannot be distinguished from k-1 other individuals
      - Temporal cloaking: reduce the frequency of temporal information
  - - - Mask an individual's precision
      - Deliberately degrade the quality of information about an individual's location (imperfect information)
      - Identity can be revealed
    - - Spatial imperfection ~~ privacy
      - The greater the imperfect knowledge about a user's lcation, the greater the user's privacy
    - - Accurate + precise
      - Inaccurate and precise
      - Accurate + imprecise
      - Inaccurate ( I not in set O ) and imprecise ( | O | > 1 )
    - - Larger the O => less info revealed about the true location I => greater level of privacy
      - Greater the distance between ) and the true location I => greater level of privacy
    - - Does not need to be contiguous
      - An individual is located at one of the vertices in O
    - - Input
        
        Weighted graph representing the geographic environment
        
        Obfuscation set O of the client's location I
        
        Points of interest
      - Output
        
        Pair <q, C>, C in set (0, 1] is the confidence that q is the nearest target to the client's current location
      - Diagram 15
      - Computationally expensive: Need to calculate the shortest path of each point of interest with each possible location
      - To save computation: connect all possible locations with a virtual place => calculates distance to restaurants from that single place => just as effective
- - - - Communication overheads
      - Security threats
      - Single point of failure
  - - - Use WPANs (spontaneous local ad-hoc networks) e.g Bluetooth
      - Clique: form spontaneous local ad-hoc networks that are wirelessly connected
      - Our mobile devices also have radio frequencies that can be used for local communication
      - Do not disclose your precise position to anyone (obfuscation) including your neighbors
      - Be k-anonymous to your location-based service provider
    - - Separate an agent's request (query requestor) from the agent requesting this service (query initiator)
    - - Locally cloaked area (LCA)
        
        represents imprecise location using obfuscation
        
        Periodically broadcast to neighbors
      - Globally cloaked area (GCA): uses obfuscation ; query requestor computes its obfuscated area using the LCA of its neighbors
      - Selection: uses anonymity ; query initiator randomly selects a query requestor to forward its request to the LSP
    - - Eliminate the rectangle whose edge has the greatest distance to the closest edge of the agent's rectange
      - Evaluate pairs of rectangles and compute distance for each pair, remove the rectangle that maximises the distance
      - Evaluate pairs of rectanges; remove rectangle that maximises area from the current GCA
    - - Anonymous selection algorithm
        
        Near-uniform random selection process
        
        Even from agents not in communication range
        
        Without disclosing IDs of involved agents
        
        Initiator chooses A -> A chooses B -> B is the requestor
      - Frequency computation
        
        Creates a map of hop counts that the initiator can use to route requests
      - Naive vs near-uniform selection: random selection will overburden a certain agent -> need to optimise energy cost
    - - NN queries
      - Group NN queries
- - - - Positions (but not identities) are known
      - User choice: smallest region for best QoS (not too small to get useless info, not too big for attacks)
  - - - Continuous query
      - Position (but not identities) are known
      - Persistent pseudonym during updates
  - - - Continuous query
      - Maximum possible speed
      - Same pseudonym during two consecutive updates
  - - - Continuous query
      - Continuous updates necessary to ensure result is correct
- - - - Provides a user with an option to trade the accuracy of query answers for trajectory privacy
      - Guarantees that the distance of the data object to the user's location is within a bound of the actual nearest data object's distance
      - Trade off
        
        Pay more: ask for a larger k (number of nearest data objects)
        
        Travel more: confidence level (travel at 1/cl); expand number of POI in query result
    - - Specify higher values for confidence level or number of nearest data objects or both
      - Resists the overlapping rectangleattack
      - Prevent identification of user's trajectory
      - Efficient evaluation of a kNN query with respect to a rectangle + specified confidence level
      - Can be done on your mobile device but eats up a lot of battery => use trusted 3rd party service
- - - - Displace
      - Replace
  - - - Senstitivity rank
      - Duration of a stop
      - Total duration of a trip
- - - - Find instruction sequences for shortest paths from all possible current locations to the destination
      - Select instruction sequence that most likely reaches the goal
      - Follow instruction sequence; ignore instructions that cannot be executed
      - If destination is reached then end, else repeat