Trust-Aware Cluster-Head Selection in IoT Wireless Sensor

Networks via Interval-Neutrosophic Information Fusion

Durdona Uktamova1,* Adnan Manzoor2

1 Faculty of Joint Degree, Tashkent State University of Economics, Uzbekistan

2 Faculty of Engineering, Central Asian University, Uzbekistan

Received: August 14, 2025 Revised: Accepted: November 20, 2025 ⋆ Corresponding author

ABSTRACT

Cluster-based routing extends the lifetime of an Internet-of-Things wireless sensor network (WSN), but the quality

of a cluster head (CH) depends on several mutually uncertain factors—residual energy, link quality, centrality and

behavioural trust—each measured imperfectly and reported as a range rather than a point. We model each candidate

node with an interval-neutrosophic set (INS), so every factor carries an interval of truth, indeterminacy and falsity,

and we fuse the factors with an interval-neutrosophic weighted aggregation operator into a single suitability score.

Cluster heads are then elected by a possibility-degree ranking of the fused interval scores, subject to a minimum-trust

guard. Simulation of a 200-node network shows that interval-neutrosophic CH selection extends first-node-death

time by 18–24% over a standard energy-and-distance heuristic and cuts the share of misbehaving nodes elected as

heads from 14.7% to 1.3%. The method degrades gracefully as measurement indeterminacy grows and adds only

modest per-round overhead.

Keywords: Interval-neutrosophic sets Information fusion Wireless sensor networks Cluster-head selection Trust

Energy efficiency

1. INTRODUCTION

An IoT deployment is often a battery-powered WSN whose

nodes cannot be recharged, so energy is the currency that

determines how long the network sees its environment.

Clustering—electing a subset of nodes as cluster heads that

aggregate and forward traffic—is the dominant way to spend

that currency efficiently. The recurring question is which

nodes should serve as heads, and the answer must balance

residual energy against link quality, node centrality and, increasingly,

trust: in an open IoT setting a compromised node

that is energy-rich but misbehaving is a poor and even dangerous

choice of head.

These factors are not only conflicting but uncertain, and the

uncertainty has distinct origins. Residual energy is estimated

from a battery and radio model that drifts with temperature

and hardware ageing. Link quality, whether derived from the

expected transmission count or the received-signal-strength

indicator, fluctuates from packet to packet as interference

and multipath change. Centrality depends on a neighbour set

that itself changes as nodes sleep and die. Trust is the most

uncertain of all: it is inferred from a finite, noisy history of

watchdog observations, so a node that has simply not been

observed much looks different from one observed often and

found honest. A single crisp score papers over all of this and,

worse, treats a confidently-good node and a barely-observed

node identically when their point scores happen to coincide.

We argue that each factor is naturally an interval with an

associated indeterminacy—narrow when the estimate is well

supported, wide when it is not— and that cluster-head se-