1 of 10

01 · RESEARCH PRESENTATION

SMART QUEUE

MANAGEMENT

SYSTEM

Machine Learning-Based Solution to Service Delivery Challenges in South Africa

Peace Khutso Molimo

Supervisor: Dr Ayinde Mohammed Usman

University of the Western Cape · 2026

Department of Computer Science

01

2 of 10

02 · PROBLEM STATEMENT

THE QUEUE

CRISIS

  • Citizens arrive before sunrise and wait 5+ hours
  • May still leave without being assisted
  • Caused by system failures, staff shortages & poor queue optimization
  • No real-time information available to citizens

5+ HOURS

Average wait time at Department of Home Affairs

63.1M

Population in 2025 — projected to exceed 65M in 2026, increasing demand

POPIA

Compliance required — sensitive citizen data must be protected

Peace Khutso Molimo · UWC · 2026

3 of 10

03 · RESEARCH AIM & SCOPE

RESEARCH

AIM

To design a machine learning-based Smart Queue Management System that addresses inefficient queue management in South African public service environments through real-time monitoring and waiting time prediction.

SPECIFIC GAPS ADDRESSED

GAP 01

Contextual Application

Limited queue management research in South African government institutions (DHA) where walk-ins & unpredictable arrivals are common.

GAP 02

Predictive Integration

Few systems combine real-time queue monitoring with ML-based waiting time prediction to proactively optimize service flow.

Peace Khutso Molimo · UWC · 2026

4 of 10

04 · LITERATURE REVIEW

KEY THEMES

01

Queue Management

Digital solutions in hospitals & government offices reduce wait times significantly. DEA models and real-time updates prove effective [1],[2],[3].

02

Predictive Techniques

ML algorithms (Random Forest, ANN, Gradient Boosting) outperform traditional M/M/s and FCFS models in complex, multi-arrival systems [4],[5],[6].

03

Multi-class Systems

Multi-class queue models outperform single-class models by 30–40% accuracy when combined with regression trees and snapshot predictors [5].

04

Cybersecurity

Public service systems in SA face ransomware, phishing & data breaches. RBAC, MFA, and staff training are key recommended mitigations [7],[8],[9].

Peace Khutso Molimo · UWC · 2026

5 of 10

05 · PROPOSED SYSTEM

THE SYSTEM

OVERVIEW

CUSTOMER

Join queue via app, web or kiosk. Track position, receive notifications, manage bookings.

STAFF

View live queue, call next customer, update service status, handle priority cases.

ADMINISTRATOR

Monitor system performance, manage users, view analytics, configure settings.

ACCESS CHANNELS

Mobile App

Web Portal

On-site Kiosk

Peace Khutso Molimo · UWC · 2026

6 of 10

06 · USER JOURNEY

HOW IT WORKS

1

Join Queue

Customer uses app, web or kiosk to register and select service type.

2

Position Assigned

System assigns queue number and records arrival time and service type.

3

ML Predicts

KNN, Random Forest & ANN models predict wait time and serving time in real time.

4

Notification

Customer is alerted when their turn approaches. Staff calls them to the counter.

5

Admin Monitors

Administrator views live dashboard queue load, staff activity and performance.

Peace Khutso Molimo · UWC · 2026

7 of 10

07 · SYSTEM ARCHITECTURE

MULTI-LAYER

ARCHITECTURE

USER INTERFACE LAYER

Mobile App · Web Portal · On-site Kiosk

APPLICATION LAYER

Queue Logic · Booking Management · Notifications

MACHINE LEARNING LAYER

KNN · Random Forest · Artificial Neural Networks

DATA LAYER

User Records · Queue History · Real-time State

SECURITY LAYER

RBAC · Multi-Factor Authentication · Activity Logging

Each layer is interconnected

The ML layer is the core; it learns from real-time and historical queue data to predict both

waiting time and serving time.

The security layer wraps the entire system ensuring POPIA compliance and data protection.

Peace Khutso Molimo · UWC · 2026

Figure 1: Proposed System Architecture

8 of 10

08 · REQUIREMENTS

SYSTEM

REQUIREMENTS

FUNCTIONAL

  • Join queue via mobile, web, or kiosk
  • Real-time queue position tracking
  • ML-based waiting & serving time prediction
  • Notifications when turn approaches
  • Staff queue management & status updates
  • Administrator dashboard & reporting

NON-FUNCTIONAL

Performance

Real-time updates with minimal delay

Security

Auth, access control, POPIA compliance

Usability

Simple for all digital literacy levels

Reliability

Consistent operation, graceful error handling

Scalability

Handles peak periods without degradation

Peace Khutso Molimo · UWC · 2026

9 of 10

9·CONCLUSION

WHAT WAS

ACHIEVED

  • Designed a multi-layer ML-based Smart Queue Management System
  • Identified two key research gaps in South African public service queue management
  • Proposed KNN, Random Forest & ANN models for waiting and serving time prediction
  • Defined functional, data, and non-functional requirements
  • Incorporated cybersecurity measures for POPIA-compliant data handling

NEXT STEPS

System implementation

Dataset collection & model training

Evaluation & testing

User acceptance testing

THANK YOU

4352859@myuwc.ac.za · University of the Western Cape

10 of 10

10 ·REFERENCES

[1] A. Yalçıner, E. Gökalp, and A. Dikici, "An Optimized Queue Management System to Improve Efficiency in Public Hospitals," International Journal of Health Care Quality Assurance, vol. 33, no. 7–8, pp. 477–495, 2022.

[5] QTech Queueing System, "Clinic Queue Management System: Real-Time Queue Updates," 2022. [Online]. Available: https://qtechqueueingsystem.com/clinic-queue-management-system/

[2] J. Tan, "Implementation of a QR-Based Clinic Queue Management System," Journal of Medical Informatics, 2025, e77297.

[3] I. Cohen, "Delay Prediction for Managing Multiclass Service Systems: An Investigation of Queueing Theory and Machine Learning Approaches," 2022. [Online]. Available: https://izackcohen.com/wp-content/uploads/2022/12/Delay_Prediction_for_Managing_Multiclass_Service_Systems_An_Investigation_of_Queueing_Theory_and_Machine_Learning_Approaches.pdf

[4] C. Xu et al., "Prediction of Waiting Times in Multi-Class Service Systems," Computers & Industrial Engineering, 2015.

[5] F. Rossi et al., "Real-Time Prediction of Waiting Times in Emergency Departments Using Machine Learning," Decision Support Systems, 2021.

[6] African Journal of Information and Communication (AJIC), "Cyber Threat Landscape in South Africa," 2023. [Online]. Available: https://ajic.wits.ac.za/article/download/12938/17798/76846

[7] BMC Medical Education, "Cybersecurity Awareness in African Healthcare Systems," 2025. [Online]. Available: https://bmcmededuc.biomedcentral.com/articles/10.1186/s12909-025-07755-x

[8] Devdiscourse, "Why South Africa's Cybersecurity Policies Struggle Without Public Trust," 2024. [Online]. Available: https://www.devdiscourse.com/article/technology/3731429-why-south-africas-cybersecurity-policies-struggle-without-public-trust