An untraceable and anonymous password authentication protocol for heterogeneous wireless sensor networks

Highlights

• An untraceable and anonymous password authentication protocol.

• Robust authentication scheme for heterogeneous wireless sensor networks.

• Low-power sensor nodes for heterogeneous wireless sensor networks.

• Protocol verification using AVISPA.

Abstract

Ensuring secure access to sensitive information in a wireless sensor networks (WSNs) remains a topic of ongoing research challenge, partly due to the wide range of potential attacks and attack vectors. In this paper, we reveal a previously unpublished vulnerability in the authentication scheme for ad hoc WSN of Chang et al.'s. Specifically, we reveal that the authentication phase of the scheme does not defend against various known attacks. We then propose a robust authentication scheme for WSNs, designed to provide security against known active and passive attacks. We then evaluate the performance of the proposed scheme using AVISPA.

Introduction

Wireless sensor networks (WSNs) typically consist of a large number of sensor nodes whose storage capacity and computing power can be fairly limited (i.e. resource constrained). Generally, these sensor nodes are placed in unattended places under the control of one or more sink/gateway nodes. Multiple sink or gateway nodes are generally used to provide a larger coverage area. Applications of WSNs include environmental monitoring, agriculture, health care, military application, disaster management, domestic and surveillance systems (Akyildiz et al., 2002), and sensor nodes include micro-controller, transceiver, external memory, analog to digital converter and power source. For example, the sensor node MICA consists of a micro-controller (Atmel ATmega 128 L, 8  MHz at 8 MIPS), a transceiver (RFM TR1000 radio 50 kbit/s), some data memory (180 + 4 kB RAM), and some external memory (512 kB Flash), and supports tiny operating system (OS). Upon request, the sensor node gathers relevant information, and processes the information prior to transmitting to the target entity such as a gateway node. The sensor network may be (i) homogenous (i.e. all nodes have the same low-power and low-storage capability), or (ii) heterogeneous (i.e. nodes vary in terms of power and storage capabilities). In this paper, we focus on heterogeneous WSNs as such networks will be harder to secure due to the diversity of configurations, etc.

In heterogeneous WSNs, information (including sensitive information) is transmitted from sensor nodes to other entity(ies) via a public channel. Such a channel can be vulnerable to attacks such as interception, insertion, deletion, and re-routing of messages. Therefore, provide secure data communication in such a setting is an active research area. One “classical” solution is user authentication and session key agreement scheme (Choo, 2009, Choo et al., 2014).

One operational challenge in enforcing security measures (e.g. user authentication and session key agreement schemes) in WSNs is the difficulty in the recharging or replacing of battery for deployed sensor nodes. Thus, when designing security measures for WSNs, one should ideally reduce the energy consumption required of the sensor node. As explained in (Chandrakasan and Heinzelman, 2000), Equations (1), (2) can be used to respectively measure the energy consumption for the transmission and receiving of messages of $l$-bits over a distance (d). Here, the free space model (fs) is used when (d) is less than a threshold value $d_0$; otherwise, the multi-path (mp) model is used.

$$E_T\left(l,d\right)=\left\{\begin{array}{cc}l{E}_{\mathit{e}\mathit{l}\mathit{e}\mathit{c}}+l{\epsilon }_{fs}{d}^2& \text{for}d<d_0\\ l{E}_{\mathit{e}\mathit{l}\mathit{e}\mathit{c}}+l{\epsilon }_{mp}{d}^4& \text{for}d\ge d_0\end{array}\right.$$

$$E_R\left(l\right)=l{E}_{\mathit{e}\mathit{l}\mathit{e}\mathit{c}}$$

In the above equations, $E_{\mathit{e}\mathit{l}\mathit{e}\mathit{c}}$ denotes the energy required by the electronic circuit, and ${\epsilon }_{fs}$ and ${\epsilon }_{mp}$ the energy required by the amplifier in free space and multi-path model, respectively.

Remark 0

Unsurprisingly, we observe that the sensor node consumes power to transmit and receive message of $l$-bits. Since consumption is directly proportional to the distance between the sensor nodes and the target entity (see Equation (3)), the minimum distance should be maintained to reduce power consumption between the entities. It is known that the gateway node has higher energy resources as well as more computing capabilities compared to the normal sensor nodes in WSNs. We denote the power consumption as $P_{consume}$ and distance between the entity as $D_{dist}$, and the equation can be defined as follows (Xu et al., 2010).

$$P_{\mathit{consume}}\propto \frac{1}{D_{\mathit{d}\mathit{i}\mathit{s}\mathit{t}}}$$

Turkanovi et al. (2014), Farash et al. (2016) and Chang and Le, (2016) proposed three authentication protocols based on the architecture described in Fig. 1. However, as the proposed protocols in (Chang and Le, 2016, Farash et al., 2016, Turkanovi et al., 2014) are designed to work in a network model with characteristics different from a WSN (e.g. the need for reduced energy consumption in deployed sensor nodes), these protocols are not suitable for WSN deployment. In (Chang and Le, 2016, Farash et al., 2016, Turkanovi et al., 2014), for example, a user directly interacts over a long distance with the sensor node and vice-versa. As pointed out in (Amin and Biswas, 2016), power consumption is directly proportional to the distance between user and sensor node. In addition, a sensor node consumes more power to transmit a message in comparison to receiving the same message. Hence, a longer distance implies more power consumption required at the sensor nodes.

Thus, in this paper, we present a modified architecture (see Fig. 2), where a user accesses (sensitive) data collected from the sensor node via a gateway node. Since the sensor nodes communicate only with the nearby gateway node, the power consumption for data communication will be significantly reduced, in comparison to existing approaches of (Chang and Le, 2016, Farash et al., 2016, Turkanovi et al., 2014).

In the literature, there have been a large number of proposed smartcard-based session key agreement protocols (Kumari et al., 2015). In 2011, for example, Das et al. (2012) and Xue et al. (2013) presented two smardcard-based user authentication and key agreement protocols for WSNs. However, it was subsequently pointed out that the protocol in (Das et al., 2012) is flawed and improved versions were proposed (Xu and Wang, 2013, Turkanovic and Holbl, 2013). Li et al. (2013), and Turkanovic and Holbl (2013) also revealed vulnerabilities in the work of Xue et al. (2013). In 2014, Turkanovic et al. (2014) presented a new user authentication and session key agreement protocol for heterogeneous WSNs using smartcard and hash function. The proposed protocol was designed to achieve lower computation cost and energy consumption, while providing user anonymity, mutual authentication and other relevant security properties in an Internet of Things (IoT) deployment. However, Ruhul et al. (Amin and Biswas, 2016) and Farash et al. demonstrated that the protocol in (Turkanovi et al., 2014) suffers from a number of security vulnerabilities and presented two improved protocols. Similarly, Chang and Le, (2016) revealed a number of vulnerabilities in the protocol of (Turkanovi et al., 2014) and presented an improved solution. However, in this paper, we reveal previously unknown security weaknesses in the protocol of Chang and Le, (2016).

We regard the contributions of this paper to be as follows:

• We reveal that the protocol of Chang and Le, (2016) is insecure against common security properties such as-off-line password guessing attack with smartcard loss, user untraceability attack, smartcard recovery attack, known session specific temporary information attack, and previous session key attack. We also demonstrate that the design of the protocol's authentication phase is flawed.

• We present a more realistic architecture for WSN (see Fig. 2), and a robust lightweight hash function authentication technique using smartcards for secure data transmission in WSNs.

We then demonstrate the security of the proposed protocol using the AVISPA tool, and evaluate the performance of the protocol.

The rest of the paper is organized as follows. In Sections 2 Revisiting Chang et al.’s protocol (, 3 Security vulnerabilities in Chang et al.’s protocol, we revisit Chang et al.’s protocol (Chang and Le, 2016) and reveal the security vulnerabilities, respectively. We outline our proposed protocol in Section 4. We then present the protocol's security verification using AVISPA in Section 5 and security analysis in Section 6. Performance evaluation is presented in Section 7. The paper is concluded in Section 8.