A Trust Evaluation Framework for Sensor Readings in Body Area Sensor Networks

Abstract

This paper addresses a framework to evaluate trustworthiness of a Body Area Sensor Networks (BASN), in particular, of sensor readings. We show that such trustworthiness is to be interpreted with respect to a certain statement or goal; its evaluation is based on quality aspects derived from observations and opinions from others. We examine relevant quality aspects of sensor readings which correspond to potential deviating behaviors of sensors. We then look at how to derive such qualities from observations taking uncertainty into the evaluation as well as decay over time. We develop an extension of subjective logic for this purpose and we show how we can compute quality properties without storing long time series. We then demonstrate this for two examples, including Galvanic Skin Response (GSR) and Electrocardiography (ECG) sensed data.

ETAS: An Efficient Trust Assessment Scheme for BANs

“…The quality of data generated and collected through sensors (i.e. data freshness without any delay) plays a vital role in decision-making for better care of the patient.There are several issues and challenges such as security [1][2][3][4][5], data quality [5] and its management [6], sensor validation [7], data consistency and freshness [8][9][10], interoperability [11][12], cost [13], transmission delay, consistent performance [14][15] and inference [16]. There exist [2,4,7-10, 17, 20, 24-26,28-29,32-33,35-40,42-43,45,47] various secure models for WBANs based on authentication, authorization, access control [11], key management [10,36,51] , and encryption but these existing models are unable to satisfy the fundamental requirement (real-time response, resource efficiency, dependability, data availability, and integrity) of body sensor networks (BSNs) since they impose extra overhead by employing heavyweight algorithmsas well as not employ multi-trust concept.…”

“…Since they assume that all the participated entities are trusted (reliable), so provide protection against only external attacks. Existing WBANs trust models [2][3][4][5][6][7][9][10][34][35][42][43] failed in terms of dependability, fast response and resource efficiency due to the incorporation of week trust function (linear, static). [17][23] [43][44] states that static trust functions with stable punishment coefficient are vulnerable to security threats.…”

“…Table 1 clearly indicates the suggested security solutions with their advantages, disadvantages and complexity analysis. [40,42] where weights can be adjusted according to patient condition [3,5,6,27]. Furthermore, weight-based TMs enforce less complexity [50][51][52][53][54][55] than other bases of trust computations.…”

“…The IEEE 802.15.6 is the most modern standard to facilitate security in WBANs. WBANs entirely rely on the recorded (monitored) information via IPSs (or biosensors) as any incorrect information about health might be dangerous for patients' life [5]. In such cases, when the sensor node (SN) itself behaves maliciously (intentionally or unintentionally due to primitive stage technology issues etc.…”