Mobility management for 6LoWPAN WSN

Abstract

When a wireless node moves across different IP domains, it needs to perform both the handover in the link layer (L2) and the handover in the network layer (L3). The L2 handover is responsible for the channel switch, and the L3 handover takes charge of the IP address change to ensure the communication correctness. The L2 handover and the L3 handover are two independent processes which are performed in series, so the total handover delay is equal to the sum of the L2 handover delay which is mainly caused by channel scanning and the L3 handover delay which is primarily incurred by care-of address (CoA) configuration. This considerable total handover latency causes severe packet loss, so it is significant to reduce the total handover latency. This paper proposes a cross-layering mobility management scheme to reduce the total handover delay and it has the following contributions: (1) This scheme combines the L3 handover with the L2 one so that they can be performed in parallel. (2) A CoA is pre-configured and the channel information is acquired in advance so that the L3 handover can be achieved without CoA configuration and the L2 handover can be performed without channel scanning. The performance of this scheme is evaluated, and the results show that this scheme effectively reduces the total handover latency.

Introduction

A wireless sensor network (WSN) is typically a multi-hop network and includes a number of sensor nodes that sense data and deliver these data for monitoring [1]. With the advent of the Internet of things and ubiquitous computing, it becomes essential to connect WSN to the Internet so that sensed data can be disseminated and accessed in real time [2], [3]. IPv6 over low-power wireless personal area networks (6LoWPAN) [4] defines a lightweight protocol stack where the link protocol is IEEE 802.15.4[5], and allows low-power networks such as WSN to use the IPv6 protocol to achieve communications with the Internet. One main feature of 6LoWPAN WSN is mobility, so it can obtain more extensive application space only if it provides good mobility support.

When a wireless node moves across different IP domains, it needs to perform both the handover in the link layer (L2) and the handover in the network layer (L3) [6]. The L2 handover is responsible for the channel switch and is made up of the channel scanning process and the association process. The L3 handover takes charge of the change of the IP address to ensure the routing correctness, and includes the care-of address (CoA) configuration process and the address binding process. In wireless networks, the L2 handover and the L3 handover are two independent processes and the L2 handover is performed before the L3 handover, so the total handover delay is equal to the sum of the L2 handover delay and the L3 one [6], as shown in Fig. 1(a). Especially, the CoA configuration latency occupies a large proportion of the L3 handover delay and the minimum latency is 1000 ms while the channel scanning delay accounts for a large part of the L2 handover latency and the minimum/maximum scanning delay is 491.5 ms/4027.2 s [4], [5], [6], [7], [8], [9]. Therefore, the total handover latency is considerable.

During the L2 handover, a node switches from one point of attachment to a new point of attachment, so the packets transmitted during this interval are inevitably lost. During the L3 handover, a node is configured with a new CoA and then the new CoA binding is performed to ensure the routing correctness. During the L3 handover process a node cannot use the old CoA to perform the communication because it has switched to a new point of attachment. Therefore, from the time when the CoA configuration starts to the time when the address binding ends, the packets transmitted are also lost. In fact, the number of packets lost is usually proportional to the total handover latency [10], so the considerable total handover latency causes the severe packet loss.

In order to improve the handover performance in 6LoWPAN WSN, some handover approaches [11], [12], [13], [14], [15], [16], [17], [18] are proposed. These approaches focus on either the L2 handover or the L3 handover to improve the total handover performance, so in these approaches the L2 handover and the L3 handover are still two independent processes and are performed in series. These approaches fail to effectively combine these two kinds of handovers to improve the total handover performance.

Therefore, we are motivated to propose a CoA pre-configuration mechanism and aim to reduce the total handover latency based on the following ideas: (1) Combine the L3 handover with the L2 one so that they can be performed in parallel. (2) A CoA is pre-configured and channel information is acquired in advance so that the L3 handover can be performed without CoA configuration and the L2 handover can be achieved without channel scanning. As a result, the L2 handover only contains the association process and the L3 handover only consists of the address binding process, so the total handover delay can be reduced, as shown in Fig. 1(b).

Based on the above ideas, this paper proposes a cross-layer mobility management scheme for 6LoWPAN WSN to reduce the total handover delay, and compared with the previous works it has the following novelties:

• 1) This scheme effectively combines the L3 handover with the L2 one so that these two handovers can be performed in parallel. As a result, the total handover delay is reduced.

• 2) A CoA can be pre-configured and the working channel information can be acquired in advance before the L2 handover. In this way, the pre-configured CoA can be used to perform the L3 handover without CoA configuration and the acquired channel information can be used to achieve the L2 handover without channel scanning, so the total handover delay is further lowered.

The remainder of this paper is organized as follows. In Section 2, the related work on mobility handover solutions is discussed. In Section 3, the 6LoWPAN WSN architecture is presented, and in Sections 4 and 5 the addressing and handover algorithms are proposed, respectively. In Sections 6 and 7 this scheme is evaluated and in Section 8 this paper concluded with a summary.