Advanced Traffic CONTROL (ATC) Signal Systems in ACTION Lab

ATC is the next generation of traffic signal standard. Although it is technically still in the draft stage, most vendors have started building and deploying it. Compared with the old NEMA and Caltrans standards, it could support much more phases and detectors within one single cabinet assembly. This is especially useful in the era of connected and automated vehicle as multiple new data must find a way to send into the controller and it could easily overwhelm the old signal control equipment. Through coupling the ATC cabinet with VISSIM, we build the first ATC cabinet-in-the-loop traffic signal simulation system (ATC-CILS) of its kind in US to provide two types of services: 1. prove the concept of pioneering traffic signal concepts and guarantee its 100% deployment; 2. Provide comprehensive training opportunities to both traffic signal engineers and technicians and this will help to fill the huge gap between traffic signal design and deployment. It will also help the agencies to estimate the training efforts if they wish to pursue the latest ATC equipment.

So far, the ACTION Lab owes or loans one Econolite ATC Cobalt Rackmount, one McCain ATCeX, one Siemens M60 and five Siemens ATC 2070 and one Intelight ATC 2070-1C and therefore we approximately cover the brands of over 85% of traffic signal market in the continental US. We also owe a fully equipped Mobotrex 332 ATC cabinet supporting four SIUs (2 outputs + 2 inputs).

Click here to download the introduction of ATC-CILS document

"Hope III " High Performance Computing Cluster

In Nov-2019, ACTION Lab purchased two GPU-enabled computing server. ACTIONLAB researchers set up the third HPC cluster and the first in University of Texas at Arlington, "HOPE III" which is composed of two computing servers equipped with total 192 CPU cores, total 500GB RAM and 16000+ GPU CUDA cores. The existing "HOPE I and HOPE II" system is also officially decommissioned from research.

Do you want to build your own HPC cluster on CentOs? Click here to download our notes

 

Do you want to build your own HPC cluster (with CUDA) on Ubuntu? Click here to download our notes

"HOPE I & HOPE II " High Performance Computing Cluster

High-performance computing clusters "HOPE" in ACTLION Lab. We should call it a "low-performance computing" instead as the computing nodes are composed of two old desktops. But in terms of structure, it contains exactly the same structure as the real HPC system, Shadow (a top 20 HPC system among the US research institutes) on our campus: CENTOS, MPI, GCC, etc. As we explored many small experimental HPC programs and it often took hours in the busy Shadow to launch our small HPC programs, we decided to build our own HPC system in the lab (bold, isn't it?)

With some assistance from our school HPC team, Slade and Lei, two Ph.D. students successfully established our own HPC system which greatly facilitate our research progress, especially the test across HPC nodes. Following the tradition of HPC community, we call it "HOPE"!

Long story short, building a HPC cluster does not need to take a long journey from earth to moon. Actually, it only takes a few steps (a few tricks as well) to get it done. Even better, you do not need to pay anything for the necessary software as everything is open-source and free. Thanks to Slade, we share our note for building your individual HPC clusters and hope it help. Welcome and enjoy the exploration of the HPC world.

In Oct-2018, Institute of Systems Engineering Research (ISER) @ MSU generously offered the ACTIONLAB a GPU-enabled computing server. Combined with our existing server, ACTIONLAB researchers set up the second HPC cluster, "HOPE II" which is composed of two computing servers equipped with total 80 CPU cores, total 250GB RAM and 4000+ GPU CUDA cores. The existing "HOPE" system is also officially decommissioned from research and relocated to the student lab for them to study and practice.