Time slot assignment for convergecast in wireless sensor networks
Introduction
A wireless sensor network (WSN) consists of autonomous devices, called sensor nodes, that are capable of wireless communication and use sensors to monitor various types of physical phenomena. Sensor nodes usually operate with batteries and are exposed to various threats such as rain or sabotage in the sensing field. It is hard to recharge the batteries of sensor nodes or to replace damaged sensor nodes. For this reason, sensor nodes are designed for low power consumption with low performance processors and low power radio modules. In general, sensor nodes are deployed to monitor physical conditions such as temperature, sound or images. Sensor nodes send monitored data to compute nodes with high performance, and these compute nodes analyze the collected data in order to make informed decisions.
Convergecast is a communication pattern used for collection of sensor data in a WSN [8]. In convergecast, sensor nodes in the network send data to a sink node using wireless communication links, possibly over multiple hops. Two types of data collection exist: (1) aggregated convergecast, in which, at each intermediate node on a multihop path, the data in individual packets are processed, typically to reduce the amount of information that has to be sent, before being packetized for the next hop, and (2) raw data convergecast, in which intermediate nodes relay data packets towards the sink node without modification [11]. Although aggregated convergecast can significantly shorten the data packets that have to be transmitted, raw data convergecast must be used with some WSN applications because intermediate nodes use low performance and low power processors that are not suitable for the required processing of data packets.
There are various WSN applications, such as video surveillance [3] and sniper localization [20], that require a guarantee on the data delivery time rather than simple throughput guarantees. A guarantee on the data delivery time is mandatory for hard real-time applications and highly desirable for soft real-time applications.
Contention-based medium access control, such as Carrier Sense Multiple Access (CSMA), is inadequate for guaranteeing data delivery time because of the possible loss of packets due to collisions and nondeterministic packets delays under high data rate conditions. On the other hand, in contention-free medium access control, such as Time Division Multiple Access (TDMA), time is divided into time slots and each sensor node transmits data only in its assigned time slots to avoid collisions. Under this scheme, a frame is defined as a period of time with a sequence of time slots that repeat in a periodic manner. The use of TDMA results in a deterministic packet delay and a guarantee on the data delivery time. Furthermore, in a wireless network, the use of fixed packet scheduling in TDMA eliminates the need for idle-time listening. This capability can be used to significantly reduce energy usage since the wireless radio modules of sensor nodes can be turned off during the time slots when no communication is assigned.
This paper focuses on raw data convergecast in a WSN. It is assumed that wireless communication in the WSN is tightly controlled. TDMA is used as the medium access protocol, and each node is only permitted to send or receive data during its allotted time slots. Data generated by a sensor node is sent to the sink node over a multi-hop network formed by wireless links. The links used during this process form a tree topology, which is referred to as a data gathering tree. This paper addresses the problem of selecting the time slots to use over each link of a data gathering tree such as to minimize the data delivery time and energy usage. This problem is formalized and analyzed, and a heuristic solution is then proposed based on the analysis. While variants of this problem have been addressed by previous researchers, the novel aspect of the proposed approach is that time synchronization and wake-up delays are added to time slots. Not only does this make the model much more practical, it also results in interesting performance and energy savings differences based on the order in which nodes go to sleep and wake up. To summarize, the main contributions of this paper are as follows.
- •
A new WSN data gathering tree model that takes into account time synchronization and wake-up delays.
- •
A theoretical lower bound on the total number of time slots required per frame.
- •
Analysis of energy savings based on the order in which time slots are used.
- •
Problem formulation for minimum data delivery time with energy conservation.
- •
A new heuristic algorithm for time slot assignment specifically designed to address the proposed problem.
Section snippets
Related work
In TDMA scheduling, time slots are assigned to avoid wireless interference. Time slot assignment algorithms assign time slots either to the nodes [19], [14] or to the wireless links [23], [7]. These algorithms are typically targeted towards minimizing the number of time slots required by each node in order to be able to communicate with all of its neighbors. In a tree topology, such as that used for raw data convergecast, sensor nodes closer to the sink node require more data transfers for
Assumptions
The following assumptions are made in this paper. First, the generated data from each node is delivered to the sink node in raw data format and cannot be aggregated by the intermediate nodes during the multihop forwarding process. If the WSN handles a large amount of data, such as is the case with high-resolution images or video, it would be hard to manipulate it quickly in the intermediate sensor nodes, which typically have low performance processors.
During the data collection phase, there is
Data delivery time
The data delivery time represents the delay from the time instant when a sensor node starts to transmit its sensor data to the time instant when the last byte of that data reaches the sink node. Since every data packet transmitted during a specific frame needs to be received by the sink node in the same frame, the maximum data delivery time of all sensor nodes in the WSN is . Hence, to reduce the maximum data delivery time of the WSN, it is necessary to make the frame size as small as
Proposed algorithm
The objectives of the algorithm are to minimize the data delivery time and energy consumed for wireless communication. We address these objectives one by one with the proposed centralized algorithm. Since it is assumed that the tree structure for data collection has already been built, the following three steps are executed using this tree.
- 1.
Data delivery time minimization
- 2.
Successive time slots rearrangement
- 3.
Notification of time slot assignment information.
Since delay sensitive data is considered,
Experimental results
To evaluate the performance of the proposed algorithm, we conducted several simulation experiments based on the model presented in Section 3. Along with the proposed algorithm, 4 variants of TRASA [1] and the algorithm proposed by Incel et al. in [11] were implemented for comparison purposes. The number of assigned time slots, the number of mode switches from sleep to active, the average duty cycle, the required buffer size and the execution time were used as performance metrics.
Conclusion
This paper has investigated time slot assignment for data collection in WSNs that operate using TDMA with the objective of minimizing data delivery time and energy consumption. Based on the concept of using empty slots in a “dominant tree”, a theoretical lower bound on the number of the time slots is derived. Moreover, a “most descendants first (MDF)” algorithm, which touches this lower bound, is proposed. Thus, both the algorithm and lower bound are the tightest solutions in practice and
Junyoung Park received the B.S. and Ph.D. degrees in Electrical Engineering from the Pohang University of Science and Technology (POSTECH), Pohang, Korea, in 2007 and 2015, respectively. He is currently a senior engineer with the System LSI Business Unit of Samsung Electronics. His current research interests include wireless sensor networks, Internet of Things, embedded systems, distributed computing, and cloud computing.
References (26)
- et al.
Link scheduling in wireless sensor networks: Distributed edge-coloring revisited
J. Parallel Distrib. Comput.
(2008) - et al.
Distributed time-optimal scheduling for convergecast in wireless sensor networks
Comput. Netw.
(2008) - et al.
A spatial correlation aware algorithm to perform efficient data collection in wireless sensor networks
Ad Hoc Netw.
(2014) - et al.
Delay optimized time slot assignment for data gathering applications in wireless sensor networks
- et al.
On tree-based convergecasting in wireless sensor networks
- et al.
Design and implementation of a real time video surveillance system with wireless sensor networks
- et al.
Scheduling for information gathering on sensor network
Wirel. Netw.
(2009) - et al.
Aggregation tree construction in sensor networks
- et al.
Tdma scheduling algorithms for wireless sensor networks
Wirel. Netw.
(2010) - et al.
Adaptive tdma-based clustering for critical event detection in wireless sensor networks
Collection tree protocol
Fast data collection in tree-based wireless sensor networks
IEEE Trans. Mobile Comput.
Fault-tolerant wsn time synchronization
Wirel. Sens. Netw.
Cited by (28)
Data aggregation and precedence by delay sensitivity (DAP-DS): Data transmission over wireless body sensor networks
2020, Microprocessors and MicrosystemsCitation Excerpt :Many improved methods of retransmission mechanism are developed to decrease the retransmission mechanism delay effect. Recent researches on data aggregation delay state that aggregation of numerous data packets to one packet [10,11], and during this situation, the cycle of data collection produces a single packet. Role of the parent node is the transmission of a single packet that is aggregated from every data packet which is generated by its branch node.
Design optimization of wireless sensor networks in precision agriculture using improved BQIGSA
2017, Sustainable Computing: Informatics and SystemsCitation Excerpt :Sensor nodes are usually powered by low energy batteries which have limitations in cost and weight [11,12]. Therefore, energy management and maximizing the lifetime of the WSNs are the most important concerns in designing WSNs [13–21]. In this proposed method, one of the strategies is to schedule nodes to alternate between sleeping and active modes.
QQIGSA: A quadrivalent quantum-inspired GSA and its application in optimal adaptive design of wireless sensor networks
2017, Journal of Network and Computer ApplicationsCitation Excerpt :Since they are designed for low power consumptions, they can keep sensor nodes alive for a limited time interval. Therefore, energy consumption management is one of the most major issues in the design of WSNs and strictly affects the lifetime of the network (Park et al., 2015; Carrabs et al., 2015; Rebaia et al., 2015; Ferentinos and Tsiligiridis, 2007; Ferentinos and Tsiligiridis, 2010; Krishnamachari and Ordonez, 2003; Ferentinos and Tsiligiridis, 2005; Hojjatoleslami et al., 2011; Rafsanjani et al., 2013; Gurkan, 2015; Ahmad et al., 2015). Another important challenge in WSNs is the connectivity requirements of the network, which certainly depends on the applied communication protocol.
A Reliable System-of-Systems Healthcare Monitoring Framework
2023, IEEE Access
Junyoung Park received the B.S. and Ph.D. degrees in Electrical Engineering from the Pohang University of Science and Technology (POSTECH), Pohang, Korea, in 2007 and 2015, respectively. He is currently a senior engineer with the System LSI Business Unit of Samsung Electronics. His current research interests include wireless sensor networks, Internet of Things, embedded systems, distributed computing, and cloud computing.
Sunggu Lee (M’88) received the B.S.E.E. (Hons.) degree from the University of Kansas, Lawrence, KS, USA, in 1985, and the M.S.E. and Ph.D. degrees from the University of Michigan, Ann Arbor, MI, USA, in 1987 and 1990, respectively. He was an Assistant Professor with the Department of Electrical Engineering, University of Delaware, Newark, DE, USA. From 1997 to 1998, he was a Visiting Scientist with the IBM T.J. Watson Research Center, Yorktown Heights, NY, USA, and a Visiting Researcher with the DREAM Laboratory, University of California at Irvine, Irvine, CA, USA, from 2005 to 2006. He is currently a Professor with the Department of Electrical Engineering, Pohang University of Science and Technology, Pohang, Korea. His current research interests include wireless sensor networks, cloud computing, real-time computing, parallel computing, and fault-tolerant computing.
Sungjoo Yoo (M’00) received the Ph.D. degree from Seoul National University, Seoul, Korea, in 2000. He was a Researcher with the TIMA Laboratory, Grenoble, France, from 2000 to 2004, and Samsung System LSI, Seoul, from 2004 to 2008. He is currently an Associate Professor with the Department of Computer Science and Engineering at Seoul National University. Prior to this appointment, he was an Associate Professor with the Department of Electrical Engineering, Pohang University of Science and Technology, Pohang, Korea, from 2008 to 2014. His current research interests include server/mobile software, architecture and circuit design for low power system-on-a-chip (SoC), and memory and storage hierarchy from cache, DRAM, nonvolatile RAM to solid-state disk. He was a recipient of the Best Paper Award at the International SoC Conference in 2006; and the Best Paper Award nominations at the Design Automation Conference in 2011, and the Design Automation and Test in Europe in 2002 and 2009.