Performance Analysis of Cross Layer Protocols for Wireless Sensor Networks Sachin Gajjar
Shrikant N. Pradhan
Kankar Dasgupta
Computer Science and Engineering Department, Nirma University Ahmedabad, India +91 9428413275
Computer Science and Engineering Department, Nirma University Ahmedabad, India +91 9825041254
Indian Institute of Space Science and Technology, Thiruvananthapuram, India +91 9496020001
[email protected]
[email protected]
[email protected]
ABSTRACT Present age technologies like Micro-Electro-Mechanical Systems for development of smart sensors, small transceivers and lowpriced hardware are fueling increased interest in Wireless Sensor Networks. The task of developing a generic protocol framework for optimizing Wireless Sensor Networks is challenging because limited processing capabilities, memory and power supply of sensor node make it difficult to cater requirements of versatile applications of these networks. This has forced researchers to dissect the traditional layered protocol design process. As a result cross layer protocols that attempt to exploit richer interaction among communication layers to achieve performance gains have emerged. This paper surveys, classifies, simulates using Network Simulator NS2 and analyzes well referred cross layer protocols namely Low Energy Adaptive Clustering Hierarchy, Self Organized TDMA Protocol, Flexible TDMA Protocol, Energy Efficient Fast Forwarding Protocol and D-MAC. We also identify possible risks associated with cross layer design and suggest precautionary guidelines for the same.
Categories and Subject Descriptors
researchers to dissect the layered protocol design process. As a result cross layer approaches which attempt to exploit a richer interaction among communication layers to achieve performance gains have emerged. A comprehensive definition encompassing all of existing cross layer approaches is given by Raja Jurdak [1] as: “Cross layer design with respect to a reference layered architecture is the design of algorithms, protocols, or architectures that exploit or provide a set of interlayer interactions that is a superset of the standard interfaces provided by the reference layered architecture.” This paper surveys, classifies (based on interaction among the OSI layers), simulates using Network Simulator NS2 and analyzes well referred cross layer protocols involving Medium Access Control and Network layers of the OSI protocol stack. The remainder of paper is organized as follows: Section 2 discusses features of WSN that become drivers for cross layer protocols Section 3 discusses cross layer protocols analyzed in paper, Section 4 presents qualitative and quantitative parameters used to analyze the protocols, Section 5 discusses simulation and analysis of cross layer protocols. Finally Section 6 gives conclusion of paper.
C.2 [Computer-communication networks]: Network Protocols.
General Terms Algorithms, Performance Evaluation, Network Simulator NS2.
2. DRIVERS FOR CROSS LAYER APPROACHES This section discusses specific features of WSN that become the drivers for cross layer approaches.
Keywords Cross layer Protocols, Wireless Sensor Networks, Survey, Simulation and Analysis.
1. INTRODUCTION A Wireless Sensor Network (WSN) is a network of sensors that, senses specified parameter(s) related to environment; processes data locally in a distributed manner and communicates information to central Base Station (BS). The complexities introduced by severely limited processing capabilities, memory and power supply in the sensor node at one end and the design needs of versatile applications at the other end have forced
2.1 Mobility Sensor nodes may not change their initial location, may change it due to environmental effects or may be carried by mobile entities. This may apply to only subset of nodes against all of them. Mobility causes parameter changes for the physical layer (e.g. interference signal levels), the data link layer (e.g. access schedules), the routing layer (e.g. topology change), and the transport layer (e.g. connection timeouts). A cross layer design enhances a node’s capability to manage its resources in mobile environments by exploiting the interdependencies between layers.
2.2 Wireless Transmission media (c) 2012 Association for Computing Machinery. ACM acknowledges that this contribution was authored or co-authored by an employee, contractor or affiliate of the national government of India. As such, the government of India retains a nonexclusive, royalty-free right to publish or reproduce this article, or to allow others to do so, for Government purposes only. ICACCI '12, August 03 - 05 2012, CHENNAI, India Copyright 2012 ACM 978-1-4503-1196-0/12/08…$10.00.
Wireless media posses some inherent adverse characteristics like signal attenuation with increase in distance, multipath fading, high Bit Error Rates, hidden and exposed terminal problem, spatial contention and reuse, media capture effect, transmission interference and unfairness in media access. All these lead to packet losses which further require retransmission or error control and thereby increase in energy requirements of the sensor node.
Hence, it can be seen that transmission errors, QoS requirements, and energy consumption are closely related. To address these problems simultaneously at all layers requires cross layer solutions [2].
2.3 Size, Resources, Energy Depending on application requirements size of a single sensor node may vary from size of a brick to a dust particle. Varying size and cost constraint result in varying limits on energy, processing, storage and communication resources. Nodes may have limited stored power source or may replenish it from environment (eg. solar cells). Cross layer design approaches can expose power and computation related variables at several layers, enabling nodes to efficiently utilize their energy and computational resources. Each layer can see a bigger picture of the system and make better decisions.
2.4 Application Specific Structure The performance requirements of application vary greatly between different sensor network applications. For example, a sensor network for environmental monitoring prioritizes network lifetime to avoid power replenishment. In contrast, a sensor network for intrusion detection system emphasizes reliable and timely delivery. Cross layer can provide application specific performance requirements.
2.5 Network Coverage and QoS Network Coverage is defined as ratio of monitored space to entire space. In case of dense and redundant deployment transmitting repeated data, results in increase in contention (MAC layer); congestion, and complex routing (Network layer) thereby resulting in waste of energy (Physical layer). In a sparse deployment, more energy will be required to reach intermediate nodes and BS, resulting in network partitioning and decrease in network lifetime. To provide real time; secured and private communication protocols should offer services such as good coverage; congestion control; active buffer monitoring; acknowledgements; message cryptography; message priority and packet loss recovery. All these require co-operation and information sharing among the layers which is possible through cross layering.
2.6 Time Synchronization Sensor’s hardware clock reference can directly affect an application operation. For instance, a WSN for target tracking is useless if it cannot register both position and detection time of an event. Thus application layer depends on physical layer. The complexity of time synchronization protocol can directly affect network lifetime. This strong interdependence among layers of protocol stack in WSN is the major driver for using cross layer approaches for protocol design in WSN.
3. CROSS LAYER PROTOCOLS FOR WIRELESS SENSOR NETWORKS
new unidirectional or bidirectional interfaces. Richer interaction enables a closer coordination between layers and optimizes the protocol’s performance. Alternatively, a cross layer based model may support comprehensive state variables accessible to all layers. In both the cases the layered architecture is intact which enables interoperability with layered architectures. Cross layer protocols based on data sharing are discussed next.
3.1.1 D-MAC D-MAC [3] enables uninterrupted data forwarding from several sources to BS on a multihop path by staggering node’s schedule wakening them up sequentially. Node’s schedule is divided into three periods: receiving (node receives a packet and sends an ACK), sending (node forwards a packet to its next hop and receive an ACK), sleeping (node turns off its radio). The receiving and sending periods have identical length μ which is long enough for transmitting and receiving one packet. If d is the depth of node in data gathering tree, node sets its wake up schedule dμ ahead from BS schedule and periodically goes into receiving, sending and sleeping states. Thus when there is no collision, packet is sequentially forwarded along multihop path to BS without sleep latency. The media access combined with packet routing decreases latency and energy spent for packets to travel from several sources to BS. The duty cycle is adaptive to traffic variation. The draw back is DMAC is designed for tree based multihop topology and does not consider the node fairness. Further, interference between nodes in same depth is to be handled carefully and MTS (More To Send) messages to adjust duty cycle under the interference add to protocol overhead.
3.1.2 Energy Efficient and Fast Forwarding Energy Efficient and Fast Forwarding (EEFF) [4] is an asynchronous MAC protocol, coupled with dynamic minimum hop routing selection. The nodes are not synchronized and perform their active-sleep schedule individually. The data is send without any preambles. When a node wants to send data, it senses the channel for time interval Tsense. If channel is free it broadcasts RTS (Request To Send) packet containing receiver’s address. Nodes receiving RTS decide whether to accept the request according to its local state (active or sleep state and hop latency to BS). If |Si| is set of sender i’s parents, DSi is average latency from i’s parents to sink plus the current hop latency, DSi’ is single hop latency from node i to any node of set S then receiver j (j Si) runs a decision function F(Si, j) to decide whether to accept the RTS packets from node i. F < 0 means average latency would be higher if node j takes part in transmission and hence it would not send CTS.
D S
i
0 1 S i
In this category of cross layer protocols adjacent or non adjacent communication layers of the model (eg. OSI) share data through
D D , otherwise S S jS j i i
F ( Si , j ) DS
i
This section categorizes cross layer protocols and introduces the cross layer protocols simulated in the paper.
3.1 Data sharing
, i is the sink
F (S , J ) i
{J }
D
Si
0,
Accept,
0,
Reject.
, L 1, 0 i T D sleep where Li = level of node i S , L 1 i i S 1 i
Candidate receivers compete to send CTS (Clear To Send) packet. The winner node sends CTS and keeps awake to receive data packet from the sender. Nodes broadcast their local routing information periodically to build minimum hop routing table information. Levels of node are decided based on the number of hops from BS. Nodes at a higher level than sender (closer to BS), will wake up first and hence will be the candidate receivers. EEFF achieves good latency performance and saves energy as dynamic routing approach selects receiver during but not before the transmission. The latency estimation does not adapt to varying network condition, routing protocol is limited to the hop-to-sink metric and does not consider traffic load in the network.
3.1.3 FlexiTP FlexiTP [5] utilizes cross layer interaction between MAC and routing layers to provide efficient methods for allocating energy, bandwidth and memory resources in WSNs. The sensor node schedule at MAC layer enables routing protocol to collect and disseminate data to and from nodes in WSNs. The two main phases of FlexiTP are: Initial network setup and Data gathering cycles. FlexiTP uses CSMA/CA to build a data gathering tree and assigns nodes’ schedules which are maintained throughout their lifetime. To avoid collision a token passing scheme is used and nodes perform procedures only if they hold a token. At the end of initial network setup, nodes perform regular data gathering tasks using their TDMA schedules. First Slot is Fault Tolerance Slot (FTS), a short CSMA period where all nodes are in listen mode allowing the nodes to adjust themselves according to their neighborhood. Nodes switch to receive mode for their scheduled receive slot list (RSL) slots, transmit mode for scheduled Transmit Slot List (TSL) slots, or else to sleep mode. A Multi Function Slot (MFS) is used for local time synchronization and local repairs. The Conflict Slot List (CSL) records slots that are used by a node’s first and second level neighbors to avoid slot conflict between neighbors. Nodes maintain Global Highest Slot (GHS) field containing network’s highest slot number. A data gathering cycle ranges from slot number one to GHS. Nodes determine their positions in time from the start and the end of a data gathering cycle. Nodes forward packets to their parents until these parents become unreachable either due to energy depletion or external environmental factors. During the data gathering cycle, a transmit slot of a node overlaps with receive slot of node’s parent. Hence only sensor nodes involved in a communication awake at a particular slot resulting in energy savings. FlexiTP becomes fault tolerant and energy efficient as nodes can build, modify, or extend their scheduled number of slots during execution based on local information. The number of children is not restricted to optimal value and children selection is based on simple broadcast reply mechanism which may lead to improper parent child pair.
3.2 Design merger In the extreme case cross layer approaches can partially or completely merge functionality of layers. Depending on the level of integration the merging of layers may result in monolithic
design. Since layering is not maintained the architecture becomes complex and is difficult to maintain and evolve when new requirements arise. Cross layer protocols based on design merger are discussed next.
3.2.1 Low Energy Adaptive Clustering Hierarchy Low Energy Adaptive Clustering Hierarchy (LEACH) [6] combines energy efficient cluster based routing and media access together with application specific data aggregation. The operation of LEACH is divided into rounds consisting of: set up phase where cluster head (CH) selection and cluster formation is done and steady state phase where nodes send their data to CH which in turn sends it to BS. Clusters are formed by using a distributed algorithm, where nodes make autonomous decisions whether to become a CH or not. During setup phase each node chooses a random number (RN) between 0 and 1. It calculates Threshold T(n) as given below. P ,if n G 1 T ( n ) 1 P * ( r mod ) P ,otherwise 0
Where P=Desired % of nodes which are CH, r=Current Round, G=Set of node that has not been CH in past 1/P rounds. If RN
