PDSLAB Research
The internet connects millions of computers together. Applications that
run on multiple computers connected by the internet are called distributed systems.
The PDSLAB (parallel and Distributed Systems Laboratory) conducts research
in the areas of:
-
Combining Replication with Error Correcting Codes
We are working on a method to implement fault-tolerant
services in distributed systems based on the idea of fused state machines.
The theory of fused state machines uses a combination of coding theory
and replication to ensure efficiency as well as savings in storage and mes-
sages during normal operations. Fused state machines may incur higher
overhead during recovery from crash or Byzantine faults, but that may
be acceptable if the probability of fault is low.
For crash faults, we give an algorithm that requires
the optimal f backup state machines for tolerating f faults in the sys-
tem of n machines. For Byzantine faults, we propose an algorithm that
requires only nf + f additional state machines, as opposed to 2nf state
machines.
-
Software Fault-tolerance of Distributed Programs
How to ensure that applcations run proplerly even when one or more computers malfunction?
We are currently working on a NSF funded project in this area.
We have developed efficient techniques for
tracking dependency in distributed systems, detecting stable
and unstable predicates, controlling distributed computations, etc.
-
Software Infrastructure for the Internet Applications
How to let common users write Internet applications?
How to harness computing power of multiple computers?
We are currently working on a project funded by Texas Higher Education
Coordinating Board for developing a distributed computing platform
for applications in Chemistry (analyzing catalysts). This project
is joint with Dr. Henkelman in the Department of Chemistry.
-
Model Checking of Distributed Programs
How can one verify the
correctness of distributed programs. We have developed a tool called
TC-SPIN that verifies correctness of a distributed program without
explicit global state enumeration. We have also developed a runtime
verification tool called POTA that verifies a single execution
of a distributed program. We are currently working on a project funded
by Semiconductor Research Consortium (SRC) for verification
of concurrent hardware.
-
Distributed Debugging
How to identify faults in distributed programs?
We have developed algorithms that allow efficient obervation and
control of distributed programs. This project has been funded
by NSF.
The publications of these projects are
here.