ImplementingReal-TimeTransactionalSecurityPropertyusingTimedEditAutomata

N.Rajamanickam and R.NadarajanPSG College of Technology, Coimbatore, India

Email: nrm@mca.psgtech.ac.inhttp://dx.doi.org/10.1145/2523514.2523578

AbstractTimed edit automaton is action sequence transformer which takes a sequence ofactions as input and produces another sequence of actions according to the policyit implements. If the input action sequence obeys the policy then timed edit au-tomaton produces same actions sequence or an equivalent action sequence. If theinput action sequence does not obey the policy then the timed edit automaton pro-duces modified action sequence. Timed edit automaton is suitable for implementingreal-time transactional security property.

Timed Edit AutomataThe timed edit automaton is a 6-tuple (Q,A, q0, C, δ, I) where

Q - countably infinite set of control locations

A - set of actions

q0 - start control location

C - set of real valued clocks

δ : A∪{null}×Q×B(C)×U → A∪{null}×Q×2C×U is deterministic transitionfunction

I : Q→ B(C) assigns clock constraints to control locations

Here B(C) denotes set of clock constraints. U denotes set of all possible clockstatuses. null denotes no (input/output) action.

TransitionsTE-Delay is for the transition between two actions, if the timed edit automaton

is in the same control location.

TE-Suppress-Insert suppresses the current action ai and inserts the new action b

TE-Insert is a discrete transition without taking any input action.

TE-Suppress suppresses the current input action ai and inserts no input action.

TE-Null is a discrete transition without taking any input action and withoutemitting any output action.

ConclusionOur timed market polices is a real-time transactional security property. Real-timetransactional security property is a combination of security property, real timeproperty and transactional property. This timed market policy can be implementedby using timed edit automaton. Timed edit automaton is a action sequence trans-former which transforms the input actions sequence to another action sequence andthe output action sequence always obeys the policy.

Operational SemanticsDelay Transitions

u ∈ I(q) u+ d ∈ I(q)(σ, q, u) −→

dTE(σ, q, u+ d)

(TE-Delay)

Discrete Transitionsσ = ai : ti;σ

′ δ(ai, q, g, u) = (b, q′, r, u′) u ∈ g u ∈ I(q) u′ ∈ I(q′)

(σ, q, u)b−→ TE(σ′, q′, u′)

(TE-Suppress-Insert)

δ(null, q, g, u) = (c, q′, r, u′) u ∈ g u′ ∈ I(q′)(σ, q, u)

c−→ TE(σ, q′, u′)(TE-Insert)

σ = ai : ti;σ′ δ(ai, q, g, u) = (null, q′, r, u′) u ∈ g u ∈ I(q) u′ ∈ I(q′)

(σ, q, u)null−−−→ TE(σ′, q′, u′)

(TE-Suppress)

δ(null, q, g, u) = (null, q′, r, u′) u ∈ g u′ ∈ I(q′)

(σ, q, u)null−−−→ TE(σ, q′, u′)

(TE-Null)

Here g ∈ B(C) is a clock constraint is called a guard. u denotes clock statuses, mapping from C to non negative real numbers. u ∈ g to mean that the clock valuesdenoted by u satisfy the guard g. u+ d denotes the clock assignment that maps all x ∈ C to u(x) + d. [r 7→ 0]u denotes the clock assignment that maps all clocks in r to0 and agree with u for other clocks in C \ r where r ⊆ C.

Timed Market Policy

Timed market policy enables an user to pay using two gateways for single transaction. First the user buys a service for the amount of n and then he/she pays a partialpayment p using gateway G1. If he/she pays the remaining amount n− p using gateway G2 within 60 seconds then the transaction will be allowed. If he/she try to payamount other than n− p or amount n− p is not been paid within 60 seconds then this transaction will be failed.