Recommend a queueing theory/network resource?

[llstelle]

I've been looking up reviews of queueing theory textbooks and seeing their tables of contents on Amazon, but I haven't found a satisfactory one.

In particular, I'm trying to tackle a class of problems where there is a bunch of resources (for example, memory in a computer or empty seats in a train), we assume that the resources are in a "linear sequence" (incoming people on the train have to pass by the first row of seats, then the second row, and so on...) and a way with dealing with queueing the "users" of these resources (for example, computations that require memory or commuters who are boarding the train).

I get to control the way/arrival rate at which these users "enter" the queueing system, and the resources are separated in discrete uniform distribution - if a memory block is already allocated to a computation, the incoming computation cannot use that memory block and travels to the next (takes a constant time) - or even better, I send the first computation ahead in the sequence of memory blocks and then have the subsequent computations occupy "preceding" memory blocks, so they never actually have to "meet".

And what I really want to do is figure out given limited resources, what is the maximum number of computations/users I can accommodate.

I am guessing this is a network of G?/D?/1 queues or something... I don't know; I've never done queueing theory before, so I'll like to know where I can know more about problems like this. I noticed that most queueing theory texts cover Markov processes first (with Poisson arrival rates), i.e. M/M/s, but here I can actually decide the calling population and control their arrival rates to eliminate waiting time altogether, and there isn't much literature on G/D distributions...

OK, basically... can someone point me in the right direction for more literature? Thanks!

 

[llstelle]

My description above is really bad. I meant, I am interested in a queueing network text (first priority).

But the exact problem I am working on is a open network of queues in series, with some fixed number of nodes, and the arrival rates of various classes of "customers" are decision variables. And I might be maximizing a utility curve that is a nondecreasing function of the sum of arrival rates.

 

[chiro]

Hey llstelle and welcome to the forums.

One book that I have personal experience with and can recommend is a book Written by Sheldon Ross. This will help you get a good foundation for queueing problems/network theory.

https://www.amazon.com/gp/product/0123756863/?tag=pfamazon01-20

Since you mention utility theory it seems like to get more specific information for your problem (which I am unfamiliar with) would come from sources like Operations Research texts and journals or economic journals.

If you already have a utility function (which seems to be what you are saying), then I think you can just use the queueing/network theory fundamentals and apply the utility function using first principles for probability and statistics. Standard optimization techniques in combination with this should do the trick.

Usually the way we model queues and similar processes is to use a markov model of some sort. Sheldons book I think would be of great use to you.

There are probably many other free resources that would offer you what you need but unfortunately I am unaware of them.

 

[Teresa_Hubsch]

You might want to check out Queuing Theory with MATLAB and SimEvents (http://www.mathworks.com/discovery/queuing-theory.html). SimEvents is a discrete event simulation software tool used with MATLAB and Simulink for hybrid simulation. I know you were asking for books and not software, but I think it might be helpful to do simulation for modeling.