Factoring in Delays

In his classes, Rifat Sipahi often chal­lenges his engi­neering stu­dents to explore a problem that is lit­er­ally out of this world. For example, he notes that oper­a­tors in Houston con­trol NASA’s Mars Explo­ration Rover, but it takes sev­eral min­utes to transmit steering instruc­tions from Earth to the vehicle — which presents chal­lenges for sci­en­tists, who must wait to see the rover actu­ally move before sending its next command.

But, asks the assis­tant pro­fessor in North­eastern University’s Depart­ment of Mechan­ical and Indus­trial Engi­neering, what if com­plex algo­rithms could be designed that account for the time delay, enabling opti­mally effi­cient man­age­ment of com­plex processes and sys­tems, like the NASA Rover.

That is the basis of a new paper pre­senting a dynamic snap­shot of the field of con­trol sys­tems with time delays. The overview paper — fea­tured on the cover of IEEE Con­trol Sys­tems Magazine’s Feb­ruary 2011 issue — exam­ines the draw­backs and ben­e­fits delays present across a wide spec­trum of areas, including supply chain man­age­ment, traffic flow and human biology. It also notes the mas­sive finan­cial, safety and health out­comes that hinge on sta­bi­lizing sys­tems with time delay.

“There may be no way to remove the delay as a player in the game. So that’s where we start,” Sipahi says. “How can we find an intel­li­gent algo­rithm that under­stands some infor­ma­tion is delayed, and uses that infor­ma­tion in a very effec­tive way?”

Human reac­tion delays exist for dri­vers, for instance, because they must observe a reason to brake before actu­ally stomping on the pedal. These delays often lead to numerous traffic flow issues, from stop-​​and-​​go traffic to collisions.

In the supply chain man­age­ment industry, Sipahi con­tinues, a com­pany may expe­ri­ence a dra­matic spike in sales, and order another huge ship­ment of product in antic­i­pa­tion that the trend will con­tinue and profits will soar. But inherent delays between the product being man­u­fac­tured and reaching inven­tory could botch the whole plan.

“The ten­den­cies of cus­tomers can shift much faster than the speed of these prod­ucts becoming avail­able,” Sipahi says.

Sipahi has studied sys­tems with time delays for a decade, and says not all have neg­a­tive effects.
In supply chain man­age­ment, adding delays could add sta­bility to these sys­tems by allowing man­agers more time to observe market trends, which would ulti­mately aid decision-​​making. In biology, algo­rithms could help researchers dis­cover inherent delays in how tumors and other dis­eases grow and spread, which could lead to the devel­op­ment of new treatments.

Sipahi was the paper’s lead author, and col­lab­o­rated with researchers from the Uni­ver­sity of New Mexico, Southern Illi­nois University-​​Edwardsville and insti­tu­tions in France and Belgium.

View selected pub­li­ca­tions of Rifat Sipahi in IRis, Northeastern’s dig­ital archive.

View the article here.

Related Faculty: Rifat Sipahi

Related Departments:Mechanical & Industrial Engineering