How rewarding better consumer choices could advance next-gen queueing platforms

An error occurred while preparing your download

How rewarding better consumer choices could advance next-gen queueing platforms

Improving how crowdsourced information is shared could curb long lines caused by inefficient human queuing behavior, a new study suggests.

Improving how crowdsourced information is shared across mobile platforms by incorporating a user penalty-and-reward system could curb long lines caused by inefficient human queuing behavior, a new study suggests.

In environments where it is vital for customers to be aware of service information, such as in restaurants, amusement parks or for transportation routes, accurate congestion information can provide real-time data about aspects like service availability and queue length.

Yet because congestion information can quickly become outdated, interruptions in queuing systems often cause users to seek other options. While such choices may serve them better individually, this behavior can make the entire system inefficient, said Ness Shroff, senior author of the study and a professor of computer science and engineering at The Ohio State University.

"If information is outdated and thus everybody's joining what appears to be the shortest path, you're going to create congestion over that path," said Shroff. This bottleneck can lead to gaps in fresh information for future customers to access and use, and impede overall service progress over time.

To better regulate this information learning, researchers have developed a way to incentivize people to choose less popular service alternatives. The proposed method is a side-payment mechanism that would periodically charge customers who contribute to overcrowding by making "selfish" choices and reward others for exploring alternative avenues.

In experiments using real-world datasets, the team found that this system was adept at balancing congestion with addressing user needs via alternative routes, resulting in steady performance. According to Shroff, adding incentivized settings to mobile queuing platforms goes a long way to making these complex systems work more sensibly for everyone.

"We calculate when the public value of fresh information is worth the congestion it takes to get it, and then build incentives that steer individual choices towards that balance," he said. "Giving incentives for people to try out different routes might in fact create better opportunities for all."

The study was published in the journal IEEE/ACM Transactions on Networking.

According to the study, this team's work is the first to examine how human choice can impact system outcomes. Hongbo Li, lead author of the study and a postdoctoral scholar at the AI-EDGE Institute at Ohio State, calls this phenomenon human-in-loop learning (HILL), noting that leveraging it can provide researchers with new insights into the growing class of service systems that rely on decentralized, customer-driven data.

"Designing a mechanism to change a user's decision to be both consistent with social welfare and long-term utility can be difficult," he said. "It has to be done in a way that doesn't directly hurt their service benefit."

A promising use-case scenario could look like this: A user visiting a car-charging station might be rewarded for choosing a less crowded location farther away, but penalized for visiting a closer station that is already at risk of becoming overloaded. Although both visits generate useful information for the operating system, the former is more valuable because curbing congestion helps reduce system inefficiencies, said Li.

"In testing, we saw that even average use saves costs and energy," he said. "This means our approach is amazingly good for the social optimum."

Besides keeping these systems more accurate, this team's mechanism would also limit expenses by using the money earned from those penalized to pay out rewards. With millions of people relying on queuing systems to navigate their day-to-day lives, these meaningful findings could inform future network design for a wide number of technologies and industries, the researchers say.

To advance the work, the team next aims to test how well their system works when people make different, unexpected choices regarding prices, risks and personal convenience.

"Our next step may be to develop mechanisms that are more robust to heterogeneous users and to test them experimentally in different scenarios," said Li. "It's important to consider human behavior in engineering, and our goal was to show that."

Other co-authors include Lingjie Duan from the Singapore University of Technology and Design.

More Ohio State News