Bonis98 / Distributed-query

Cybersecurity master's thesis project

Assignment

This is a project made for the master's thesis in my cybersecurity degree. The project consists in the implementation of a variant of the algorithm proposed in Distributed Query Execution under Access Restrictions

Installation and run

1. Clone this Repo
2. cd into the project root folder, and run python3.9 -m venv env to create a virtual env
3. Run source env/bin/activate to activate the virtual env
4. Run pip install -r requirements.txt to install all the packages needed to run the project
   • If you don't have pip installed, you can find informations here
5. The script has five command line arguments:
   • -p PATH, --path PATH: representing the path where to save the pdf containing the tree resulting from the computation (e.g. '../' to save the pdf in the directory containing the script folder)
   • -m ASSIGNMENT, --manual ASSIGNMENT: Manually assign node to candidate, in the form 'XYZ' to assign them to nodes in pre-order visit of the query tree plan
   • -i INPUT, --input INPUT: Path from where take the input of the algorithm
   • -v, --verbose: Enables verbose logging
   • -d, --debug: Enables debugging loggin

back

Input data to algorithm

Inputs to the algorithm are given by four different CSV files:

• relations.cvs
• tree.cvs
• subjects.cvs
• authorizations.cvs

Folder CSV_data contains an example of that files

relations.csv

This is the file modeling the base relations of the query, structured as follows:

• name: Name of the base relation
• primary_key: Primary key of the relation
• provider: Storage provider storing the relationship
• plain_attr: Attributes of the relationship stored in plain text
• enc_attr: Attributes of the relationship stored encrypted
• attr: All the attributes of the relationship (plain and enc)
• enc_costs: list of (semi colon separated) encryption costs of attributes in the relationship
• dec_costs: list of (semi colon separated) decryption costs of attributes in the relationship
• size: list of (semi colon separated) attributes sizes (used to estimate computational cost of node)
• node_id: id of the leaf node in the tree to which associate the base relation (see nodes.csv)

Parsing of relations.csv produces the following two base relations:

1. relation flight(NDPC) assigned to storage provider F
   • N is stored encrypted, it has an encryption cost of 1, a decryption cost of 4 and a size of 7
   • D is stored in plain text, it has an encryption cost of 2, a decryption cost of 5 and a size of 8
   • P is stored in plain text, it has an encryption cost of 3, a decryption cost of 6 and a size of 9
   • C is stored encrypted, it has an encryption cost of 4, a decryption cost of 7 and a size of 9
2. relation company(SJI) assigned to storage provider C
   • S is stored encrypted, it has an encryption cost of 1, a decryption cost of 4 and a size of 7
   • J is stored in plain text, it has an encryption cost of 2, a decryption cost of 5 and a size of 8
   • I is stored encrypted, it has an encryption cost of 3, a decryption cost of 6 and a size of 9

tree.csv

This is the file modeling the query tree plan, structured as follows:

• ID: Used to associate a relationship with a leaf node
• operation: Operation of the query associated with the node
  • Projection
  • Selection
  • Cartesian
    • Cartesian product requires to put manually attributes in Ap, Ae and enc_attr taken from children
  • Join
  • Group-by
  • Encryption
  • Decryption
  • Re-encryption
• Ap: Set of attributes that need to be in plaintext to evaluate the operation associated with the node
• Ae: Set of attributes that need to be re-encrypted to evaluate the operation associated with the node
• As: Remaining set of attributes
• print_label: label of the node to print when tree is exported
• group_attr: if the operation associated with the node is a group-by, this is the set of attributes on which the group-by clause is evaluated
• parent: parent node of current node, used to build the tree

Parsing of tree.csv produces the following tree:

subjects.csv

This is the file modeling the subjects involved in query computation with its authorizations, structured as follows:

• subject: Name of the subject
• comp_price: computational price of the subject
• transfer_price: transfer price of the subject

Parsing of subjects.csv produces the following subjects:

• U with computational price 1 and transfer price 1
• X with computational price 2 and transfer price 2
• Y with computational price 3 and transfer price 3
• Z with computational price 4 and transfer price 4
• F with computational price 5 and transfer price 6
• C with computational price 6 and transfer price 7

authorizations.csv

This is the file modeling the authorizations involved in query computation, structured as follows:

• subject: Name of the subject (already specified in subjects.csv)
• plain: Attributes for which the subject is authorized to view in plaintext form
• enc: Attributes for which the subject is authorized to view in encrypted form

Parsing of authorizations.csv produces the following authorizations:

• [NCPSJI,-]→U
• [PC,NSJI]→X
• [DPJI,CNS]→Y
• [NCS,PJI]→Z
• [-,NDPCJI]→F
• [-,NCSJI]→C

back