Practical packet combining for use with cooperative and non-cooperative ARQ schemes in resource-constrained wireless sensor networks
Introduction
To help address the issue of energy consumption in WSNs this paper presents a highly efficient implementation of a simple but practical technique referred to as post-detection packet combining.1 Packet combining can be used when the same data is received from multiple sources or from the same source over different time slots. The former case generally occurs in a promising technique referred to as cooperative communications, which has been shown to be of benefit for wireless networks [2] and is a viable alternative to Multiple-Input, Multiple-Output (MIMO) techniques which are not fully realisable on current WSNs due to their size, cost, and hardware limitations. The latter case can be simply referred to as a traditional Automatic Repeat Request (ARQ) system with packet combining [1].
Combining techniques such as Maximal-Ratio Combining (MRC) [3] or the use of Space Time Block Codes (STBCs) [4] are not viable options for commercially available architectures such as the popular Tmote Sky [5], MicaZ [6], and other similar devices, as their simplified hardware does not permit, for example, access to the analog portion of the transceiver. In addition, Channel State Information (CSI) at both the transmitter and receiver is generally not available and only partial information can be obtained at the receiver in most cases (in the form of Link Quality Indicator (LQI) and/or Received Signal Strength Indicator (RSSI)). As it is towards these low power, resource-constrained architectures that the current work is aimed, this paper therefore focuses on a post-detection (after demodulation) combining scheme rather than a pre-detection (before demodulation) combining scheme.
Although significant advances are likely to be made in the area of chipset design in the coming years the size of future WSNs are envisaged to become increasingly small in size (the original intention being on the order of a piece of dust [7]). It is reasonable to expect that there will remain a need for less complex protocols, despite these advances in CPU speed, available program memory, etc. Toward this end an efficient packet combining implementation referred to as improved Packet Merging (iPM), a preliminary outline of which was provided by the authors in [8], is extended and thoroughly analysed in this paper. The analysis is carried out using both simulations and practical experiments on a commercially available mote platform. The results clearly demonstrate both the significant energy savings of iPM and the beneficial effects of the underlying packet combining scheme on a real system. The packet combining procedure used in iPM is based upon previous work in [9], [10] and most recently [11]. The iPM algorithm uses available metrics such as RSSI/LQI combined with other novel techniques to significantly reduce the required complexity of the packet combining procedure.
The aim of iPM is that it can be easily implemented on low-power, resource-constrained wireless sensor nodes. The purpose of such a scheme is to reduce the required energy consumption of a wireless node while maintaining the same performance levels or, alternatively, to obtain improved performance at the same energy consumption. As this paper is concerned with WSNs, and their inherent energy limitations, it is the former case that is of interest. The combining scheme is implemented in both a cooperative and non-cooperative setup to determine the gains possible; this is done both practically and through simulations. Although the concept of packet combining is not new, very few publications demonstrate practical evidence of its benefits and instead only document theoretical results. The aim of this work is not to make any new theoretical contributions to the field but to focus solely on the practical aspects and efficient implementation of previously proposed theoretical results. With this in mind the following contributions are made in this paper:
- •
A highly efficient implementation of a post-detection packet combining scheme based on the Cyclic Redundancy Check is developed. The improved packet combining scheme offers more than an 85% reduction in the number of iterations required to find the correct packet over current implementations. This leads to significant energy savings.
- •
A thorough energy analysis is carried out for a commonly used platform based on the IEEE 802.15.4 standard to help quantify these savings.
- •
The use of a second generator polynomial for the Cyclic Redundancy Check is shown to be of benefit in this scheme offering improved performance through the significant reduction of false-positives.
- •
The packet combining scheme is shown through simulations and practical experiments to have a positive effect on the energy consumption of a wireless node as it lowers the number of transmissions required to achieve a desired packet error rate. The experiments are carried out for both cooperative and non-cooperative scenarios, and for different packet sizes.
The next section discusses background material and relevant work. Section 3 then discusses general packet combining and analyses the gains achievable through more efficient combining algorithms using simulations. Section 4 reviews the preliminary improved version (iPM) developed by the authors in [8]. Section 5 discusses an initial experimental setup used in both [8] and the current work. Packet combining issues, such as hidden errors and false-positives, are studied in Section 6 with the help of the experimental results. Section 7 looks at further improvements to iPM and provides experimental results to show the benefits of the proposed techniques. Section 8 looks at energy measurements carried out on a popular IEEE 802.15.4-compliant wireless mote. Section 9 carries out experiments to obtain PER curves on the wireless motes and shows the benefits of it on a real system. Finally, Section 10 concludes the paper and discusses future work.
Section snippets
Background and related work
Packet combining was first described by Sindhu [9] using forward error correction in bursty channels. An incremental redundancy combining technique called code combining was then developed by Chase [12] and was designed with very noisy channels in mind. Code combining concatenates the received packets to form codewords from increasingly longer and lower-rate codes. In [13] a scheme is described whereby the transmitter sends a packet to L different nodes within a cluster. Code combining is then
Packet merging
This section discusses the packet merging procedure and simulation results pertaining to the gains achievable using it. It should be emphasised that packet merging is a post-detection combining scheme. Modern transceivers used in WSNs generally cannot implement pre-detection combining schemes such as MRC as no access to the analog portion of the receiver is given. The theoretical underpinnings of packet merging are examined in particular in [10] but also in [15], [16], [17]. A receiver decision
Improved Packet Merging (iPM)
The authors developed two improved packet merging algorithms in [8], where preliminary results were presented. The first algorithm, referred to as Simple Packet Merging (sPM), chooses the same packet as mch on each run of the algorithm possibly based on some a priori topological information. As there are a total of possible error candidates the algorithm runs through each possible candidate (ec) from 1 to in unit increments, referred to as an “incremental search”, modifying mch
Experimental setup
The experiments carried out in [8] were extended in this work to examine the new improvements to the algorithm (to be described in Sections 7.2 Parity equivalence, 7.1 Use of two generator polynomials). In this section the experimental setup used in both [8] and the current work is described in more detail. The setup used a three node architecture: a Source (S), a Relay (R) and a Destination (D). The particular devices used were the popular Tmote Sky motes [5] which incorporates a 2.4 GHz
Packet merging issues
As with all error correction schemes there is a limit to the amount of errors that can be corrected. This section looks at the concepts of hidden errors and false-positives. While a method is described for dealing with false positives, the concept of hidden errors is only briefly touched upon here but is more thoroughly dealt with in [10], [11].
Further iPM enhancements
This section proposes further enhancements to iPM which improve both its efficiency and robustness. These include a reduction in the number of required iterations by one-half using a technique referred to as parity equivalence, and the effective elimination of false-positives using two distinct generator polynomials for the original and retransmitted packets.
Energy measurements
Using iPM the overall energy consumption of the network can be reduced, not only through a decreased number of retransmissions but also through a reduction in the number of collisions and failed transmission attempts (in the case of a carrier sense protocol). To help determine the performance of iPM this section studies the energy consumption of a typical mote type architecture and discusses the energy cost of performing a merging operation; this is then compared with communication costs. The
Practical results
An experiment was setup up in order to determine the error-correction performance of iPM on a real system. Each of the four situations simulated in Section 3.2.1 were considered (i.e. tradARQ, tradARQC, coopARQ, coopARQC) and the value of w(VA)max set to values of 10 and 15, as was emphasised for the simulations. A number of issues existed in trying to obtain PER curves in the practical experiments. Most significantly, it was very difficult to obtain an accurate estimation of the received SNR.
Discussion
A highly improved implementation of a simple packet combining scheme referred to as improved Packet Merging (iPM) has been developed in this paper. It has been shown that even though complex combining schemes such as maximal-ratio combining are currently not possible on resource-constrained wireless sensor networks, it is still possible to achieve quite significant performance gains using the simple techniques proposed. This section concludes the work carried out in this paper and provides some
Damien O’Rourke graduated from Dublin City University (DCU) in 2000. Following a short period as a network design engineer he went on to pursue a research masters degree in attacking embedded cryptosystems. Upon graduating in 2003 he took on a full-time lecturing position which later led him to pursue his PhD in Wireless Sensor Networks which he obtained in 2009. He has been working as a post-doctoral research fellow for CSIRO since October 2008 and his current research interests include the
References (33)
- et al.
Efficient ARQ protocols for exploiting cooperative relaying in wireless sensor networks
Computer Communications
(2007) Error Control Systems for Digital Communication and Storage
(1994)- et al.
Cooperative communication in wireless networks
IEEE Communications Magazine
(2004) Wireless Communications
(2005)- et al.
Space-time codes in wireless communications
IEEE Potentials
(2003) - Moteiv Corporation, Tmote Sky Datasheet. <http://www.moteiv.com> (accessed...
- Crossbow Technology Inc. <http://www.xbow.com/> (accessed...
- J. Kahn, R. Katz, K. Pister, Next century challenges: mobile networking for smart dust, in: MobiCom ‘99: Proceedings of...
- D. O’Rourke, C. Brennan, A practical implementation of an improved packet combining scheme for wireless sensor...
Retransmission error control with memory
IEEE Transactions on Communications
(1977)
An ARQ scheme with packet combining
IEEE Communications Letters
Code combining – a maximum-likelihood decoding approach for combining an arbitrary number of noisy packets
IEEE Transactions on Communications
An adaptive ARQ scheme with packet combining for time varying channels
IEEE Communication Letters
Cited by (5)
ReLoRaWAN: Reliable data delivery in LoRaWAN networks with multiple gateways
2023, Ad Hoc NetworksReliable and prioritized communication using polarization diversity for industrial Internet of Things
2016, 2016 IEEE Conference on Wireless Sensors, ICWiSE 2016Interference-aware relay assignment scheme for multi-hop wireless networks
2015, Wireless NetworksAn experimental study of selective cooperative relaying in industrial wireless sensor networks
2014, IEEE Transactions on Industrial InformaticsA novel CARQ algorithm for TDMA-based single source single relay wireless networks
2014, Applied Mechanics and Materials
Damien O’Rourke graduated from Dublin City University (DCU) in 2000. Following a short period as a network design engineer he went on to pursue a research masters degree in attacking embedded cryptosystems. Upon graduating in 2003 he took on a full-time lecturing position which later led him to pursue his PhD in Wireless Sensor Networks which he obtained in 2009. He has been working as a post-doctoral research fellow for CSIRO since October 2008 and his current research interests include the development of efficient event detection and signal processing algorithms for Wireless Multimedia Sensor Networks, as well as diversity techniques for Wireless Sensor Networks in general.
Conor Brennan graduated from Trinity College Dublin (TCD) with a BA (Mod) in Mathematics in 1994 and a PhD in 1998. He spent several years as a post-doctoral researcher in Dr. Peter Cullen’s wave scattering group in TCD before being appointed lecturer in the School of Electronic Engineering, Dublin City University in 2003. His research interests are in numerical methods for wave propagation modelling as well as energy-efficient communications. He has written over 50 peer-reviewed papers in these areas. He is a member of the IEEE and the IET as well as the Royal Irish Academy Committee on Communications and Radio Science.