A power-aware MAC layer protocol for real-time communication in wireless embedded systems
Introduction
In the recent years, the interest on networked wireless systems has experienced an exponential growth, mainly due to the wide range of applications, including defense systems, health monitoring, domotics, intelligent buildings and industrial control systems.
The delay introduced by the network has a significant impact on the system performance, which can be specified according to different Quality of Service (QoS) levels. For example, time-critical data related to alarms must be delivered within stringent deadlines, and control loops data have to be transmitted periodically with a bounded delay variation (jitter).
When considering the design of a communication stack for real-time systems, a deterministic Medium Access Control (MAC) layer is crucial to guarantee a bounded transmission delay for any packet sent throughout the network. The techniques adopted to handle the channel access can be roughly divided in three categories: contention based, scheduling based, and hybrid approaches. The former makes use of CSMA/CA or ALOHA (Rappaport, 1996) methods, the second one implements scheduling algorithms to rule the channel access and the latter is a combination of both, see for instance IEEE 802.15.4 Std-2011 (2011). Each approach has its own advantages and drawbacks. CSMA/CA is simple, robust, highly scalable and does not need clock synchronization between nodes. The downside is that it suffers access collisions where two or more nodes can access the channel at the same time, causing a delay in the message transmission. Moreover, since carrier sensing does not work for nodes more than one hop away, a handshake mechanism (The IEEE, 1999) is necessary to mitigate the hidden/exposed terminal problem (Tobagi and Kleinrock, 1975). As a consequence of both collisions and handshake, the network throughput can be greatly reduced. On the other hand, scheduling based methods do not suffer hidden/exposed node problems, are collision-free and highly predictable in terms of transmission delay. The main shortcoming is, in many cases, the need of some form of clock synchronization between nodes that increases the protocol overhead; the network scalability is more difficult to achieve and much more infrastructure support is needed with respect to CSMA/CA.
In battery-operated systems, the energy management represents another key issue to be addressed at design time. As highlighted by Ye et al. (2004), four main sources of energy waste can be identified at the MAC level: Overhearing, this is the energy wasted by a node when receiving packets directed to other nodes; Collision, if a packet is corrupted, it has to be resent, hence both the sender and the receiver have to consume additional energy to exchange the packet; Control Packet Overhead, this is the energy consumed by a node to send and receive control packets; and Idle Listening, this is the energy dissipated by a node in receiving mode while waiting for incoming messages.
Note that CSMA/CA protocols are particularly prone to collisions and idle listening, whereas scheduling based protocols are mainly affected by control packet overhead.
Radio devices available in the market have different operating modes, each characterized by a different level of power consumption. The most common are: sleep, receiving, and transmitting. In this work, the possibility offered by the sleep mode is exploited to reduce energy consumption.
This paper describes the Wireless Budget Sharing Token (WBuST) protocol, which is a MAC layer protocol designed for real-time wireless sensor/actuator networks (WSNs) of embedded devices. WBuST can handle both real-time and best effort traffic in multi-hop networks, while saving energy to guarantee a desired lifetime. The channel access is handled by a mixed approach that adopts a bandwidth-reservation mechanism to guarantee the desired performance, and a contention-based mechanism for the transmission of control and management messages.
Network devices are grouped into clusters of adjacent nodes, and a different radio channel is assigned to each cluster. In this way, the transmissions within adjacent clusters can take place at the same time without interfering with each other. The clusters can be connected to form various network topologies, where each cluster is managed by a coordinator, which is the node with the best link quality to neighbor nodes.
The most relevant contribution of this work is the analysis of the protocol performance, which provides a powerful method for guaranteeing a desired QoS level and lifetime for a given amount of network traffic. This is particularly useful for implementing admission control mechanisms to handle overload conditions. Concerning power management issues, besides of minimizing the energy consumption, as done in most of the related works reported in Section 2, this work also provides a method for selecting the protocol parameters that guarantee a given network lifetime. The properties of WBuST are also validated through experimental results.
This paper extends and completes a preliminary work by Franchino and Buttazzo (2012) in several directions: first, it extends the state of the art to present further related works; it proposes a methodology based on Network Calculus that can be applied to determine the maximum end-to-end inter-cluster communication delays of real-time streams in cluster-tree networks; then, it derives bandwidth and buffering requirements for network routers to guarantee the deadlines of real-time streams; and finally, it includes a new set of experimental results carried out to assess the performance of the WBuST protocol in a multi-hop scenario and show the effectiveness of the proposed energy saving mechanism.
The rest of the paper is organized as follows. Section 2 analyzes the related works, Section 3 describes the proposed protocol in detail, Section 4 introduces the traffic model, the bandwidth allocation schemes and the analysis of the protocol performance for intra-cluster communications. The method used by the protocol to save energy and predicting the network lifetime is shown in Section 5. Section 6 describes how to compute the maximum end-to-end communication delay of real-time streams, bandwidth and buffering requirements of router nodes through a technique based on Network Calculus. Section 7 reports and discusses the experimental results and, finally, Section 8 states the conclusions.
Section snippets
Related work
Real-time communication and energy saving issues over wireless networks have received great consideration in the literature during the last years. However, not many authors addressed both problems simultaneously.
Caccamo et al. (2002) proposed a cellular network architecture with a MAC protocol based on the Earliest Deadline First (EDF) algorithm (Liu and Layland, 1973). Implicit prioritization is achieved by exploiting the periodic nature of the traffic in sensor networks. The authors analyzed
The WBuST protocol
This section describes the proposed protocol in detail. WBuST is a MAC layer protocol that can operate both in single-hop and in multi-hop networks, serving different kinds of communication flows.
Budget allocation and protocol properties
This section analyzes the timing properties of the protocol in order to perform real-time guarantee tests on message deadlines. In particular, worst-case transmission times are derived for a number of bandwidth allocation schemes.
Referring to the CW structure shown in Fig. 1, it can be noted that cluster nodes access the channel, one by one, in a circular fashion, and the access time of node i is limited by the slot budget Bi. In other words, the channel access is regulated by a weighted round
Energy saving mechanism
As already mentioned, a sleep slot is allocated at the end of each CW to allow cluster nodes to turn off their radio transceiver. This section describes how to calculate the dimension of this slot to guarantee a desired lifetime for each network cluster.
Before describing how to compute the sleep slot, it is worth showing how to identify the sources of energy consumption mentioned in the introduction. Fig. 8 shows the CW of cluster C2 in the example of the cluster-tree structure described in
Multi-hop delay analysis
This section considers the inter-cluster communication and describes a method to derive the maximum end-to-end communication delay of message streams, the bandwidth and buffering requirements for the router nodes of a cluster-tree network.
Data traffic in cluster-tree networks can be formed by both upstream and downstream flows. Usually, downstream messages carry queries or control information from root to cluster nodes and upstream flows carry critical messages, e.g. sensor data, from cluster
Experimental results
The effectiveness of the WBuST protocol has been evaluated by a set of experiments carried out on a network of 10 FLEX boards (FlexBoards, 2013) equipped with a 16-bit microcontroller and a IEEE 802.15.4 compliant transceiver. The firmware has been written in C under the ERIKA Enterprise real-time kernel (ErikaEnterprise, 2013).
All experiments refer to a cluster of 9 nodes plus the coordinator. Two message streams are assigned to each node, having a total amount of 18 real-time streams. The
Conclusions and future work
This paper presented WBuST, a MAC layer protocol for time sensitive communication in wireless embedded systems. WBuST supports both real-time and best-effort traffic in multi-hop networks, grouping the network devices into clusters managed by a coordinator node. A different radio channel is assigned to each cluster, where the nodes are synchronized by the coordinator through the transmission of a periodic beacon. The channel access is regulated by a budget reservation mechanism that guarantees
References (41)
- et al.
Effectiveness of the FDDI-M protocol in supporting synchronous traffic
J. Syst. Softw.
(2001) - Abdelzaher, T.F., Prabh, K.S., Kiran, R, 2004. On real-time capacity limits of multihop wireless sensor networks. In:...
- et al.
A survey on mac protocols for wireless adhoc networks with beamforming antennas
IEEE Commun. Surv. Tutor.
(2012) - Bui, B.D., Pellizzoni, R., Caccamo, M., Cheah, C.F., Tzakis, A., 2007. Soft real-time chains for multi-hop wireless...
- Caccamo, M., Zhang, L.Y., Sha, L., Buttazzo, G., 2002. An implicit prioritized access protocol for wireless sensor...
- et al.
Time properties of the FFDI-M medium access protocol
Comput. J.
(1997) - Chen, D., Chan, E., Lee, C.-H., 1998. Timing properties of the FDDI-M medium access protocol for a class of synchronous...
- Crenshaw, T.L. Tirumala, A., Hoke, S., Caccamo, M., 2005. A robust implicit access protocol for real-time wireless...
- et al.
Mac protocols for wireless sensor networksa survey
IEEE Commun. Mag.
(2006) - ErikaEnterprise, 2013. Rtos Erika Enterprise. URL...
The evolution of mac protocols in wireless sensor networksa survey
IEEE Commun. Surv. Tutor.
Evaluation and improvement of response time bounds for real-time applications under non-pre-emptive fixed priority scheduling
Int. J. Prod. Res.
Cited by (9)
PARSECS_RT: A real-time PARSECS-based communication protocol stack for critical sensing applications
2024, Internet of Things (Netherlands)Adaptive and Beacon-based multi-channel protocol for Industrial Wireless Sensor Networks
2019, Journal of Network and Computer ApplicationsCitation Excerpt :Thus, the development of new adaptive protocols may be also useful for applications that use the mmWave spectrum. Some authors have proposed novel MAC protocols for real-time communication based on the physical layer of the IEEE 802.15.4 standard (Flammini et al., 2009; Bartolomeu et al., 2017; Franchino and Buttazzo, 2017; Correia et al., 2015). Other works proposed mechanisms to improve the performance of the MAC protocols defined by the standard (Du and Roussos, 2011, 2013; Serizawa et al., 2017; Gomes et al., 2017a; Kotsiou et al., 2017b).
Survey and systematic mapping of industrial Wireless Sensor Networks
2017, Journal of Network and Computer ApplicationsCitation Excerpt :All the end-nodes also use the same channel for communication, but the same channel can present different characteristics for different nodes, due to the spatial variations in the quality of the channel, and the diverse spatial allocation of the nodes which leads to different channel qualities. The Wireless Budget Sharing Token (WBuST) protocol is proposed in Franchino and Buttazzo (2017). The WBuST is a hybrid MAC protocol with contention-based communication and contention-free communication.
Energy preservation and network critic based channel scheduling (EPNCS) in cognitive radio sensor networks
2021, International Journal of Information Technology (Singapore)Ensuring end-to-end dependability requirements in cloud-based bluetooth low energy applications
2021, International Conference on Embedded Wireless Systems and NetworksA review on ultra reliable and low latency communications (PHY and MAC layer perspectives)
2020, Journal of Computer Science