Explain the difference between discretionary and mandatory access control.

Describe the Bell-LaPadula security policy model.
Discuss how one might implement a system enforcing the Bell-LaPadula model.

Explain what covert channels are, and how they can limit the usefulness of
multilevel secure systems.
9 Pi Calculus
Define the term sorting, and explain what it means to say that a term of the
π-calculus respects a given sorting.
Define the basic reduction rule COMM of the polyadic π-calculus. If R −→ R0
is
an instance of COMM, and R respects a sorting ob, show that R0
also respects ob.
Argue informally that this holds for any reduction R−→R0 whatever. [5 marks]
A labelled binary tree (lbt) is either a node with two sub-lbts, or else a tip with a
label (some value). Show how to represent lbts uniformly as π-calculus agents with
no free names. What sorting does your representation respect, given a sorting for
label values? [7 marks]
Translate into the π-calculus the case-switch construction
case t of ? node(t1t2) ⇒ P
? tip(v) ⇒ Q
so that, when interacting with an lbt T located at t, it will enter P (with t1, t2
bound to the locations of the sub-lbts) if T is a node, and otherwise will enter Q
(with v bound to the location of the label value). [3 marks]
3 [TURN OVER
CST.95.7.4
10 Running a Business
The Wizzo project has two phases. Each phase consists of three main tasks:
analysis, coding and test, which must be performed sequentially. Analysis for
phase two can begin immediately on the completion of analysis for phase one,
but coding for phase two must await satisfactory testing of the phase one code.
Analysis is expected to take 3 weeks for phase one, and 4 weeks for phase two,
while coding is expected to take 2 weeks for phase one and 3 weeks for phase two.
Testing is expected to take 1 week for both phases.
(a) Draw the PERT and GANTT charts for the Wizzo project, and define the
critical path. 
(b) During testing of phase one a serious bug is found resulting in an extra 2 weeks'
work required in that phase. How will this affect the overall timescale and
critical path of the project? [5 marks]
(c) An extra programmer is assigned to the project in week 6, potentially
increasing code productivity by 50%, but first requiring 1 week's training by
the project analyst. Will this allow the project to finish early, with or without
the event described in (b) above? [5 marks]
(d) What other tasks would you expect to be under the control of the Wizzo
project manager and be included in a typical software project? [5 marks]
4
CST.95.7.5
11 Specification and Verification II
Describe the choice operator, ε, of higher-order logic, giving examples of its use.
[6 marks]
Using First and Next, where
First p t = (∀t'. t' < t ⇒ ¬(p t')) ∧ p t
Next p (t1,t2) = (t1 < t2) ∧ (∀t.t1 < t ∧ t < t2 ⇒ ¬(p t)) ∧ p t2
define a function TimeOf such that for a function, f, TimeOf f n returns the time
when f becomes true for the n
th time. [4 marks]

Transcribed Image Text:

Matrix Sum Assignment Write a multithreaded function to sum all the values in an n by n matrix. Write as java that works serially but add /*spawn*/ and /*sync*/ in the appropriate places to provide multithreading. You can just complete the following java program: 1) It should be runnable and get the right answer. 2) the multithreading should not introduce race conditions 3) I don't think you can do this with parallel for. If you really think you don't need recursion and that parallel for is just fine, then /*parallel */ and /*new*/ can be introduced. But good luck! Put your analysis of your multi-threaded matrix sum into your Java file as a comment at the bottom. It should give an explanation of your reasoning. Matrix Sum Assignment Write a multithreaded function to sum all the values in an n by n matrix. Write as java that works serially but add /*spawn*/ and /*sync*/ in the appropriate places to provide multithreading. You can just complete the following java program: 1) It should be runnable and get the right answer. 2) the multithreading should not introduce race conditions 3) I don't think you can do this with parallel for. If you really think you don't need recursion and that parallel for is just fine, then /*parallel */ and /*new*/ can be introduced. But good luck! Put your analysis of your multi-threaded matrix sum into your Java file as a comment at the bottom. It should give an explanation of your reasoning.