- The resource limited nature of WSNs requires that protocols implemented on these networks be energy-efficient, scalable and distributed. In this paper, we present a novel combined routing and MAC protocol. The protocol achieves energy efficiency by minimizing signaling overhead through stateless routing decisions that are made at the receiver rather than at the sender. The protocol depends on a source node advertising its RSSI to its neighbors, which then contend to become the receiver of the packet, by measuring their local optimality index, and map this into a timer value. More optimal nodes have smaller timer values and so respond before less optimal nodes. The proposed solution is assessed through simulations. Performance results show the advantages of the proposed solution when compared to RBF protocol, a recently proposed cross-layer approach with similar goal.
In the last few years, Wireless Sensor Networks (WSN) have become a hot topic not only for researchers but also for the industry. A WSN consists of a big set of lightweight and cheap devices with many integrated sensors and wireless communication interfaces. Nodes use their wireless radio to communicate the information acquired with their sensors. When the destination is out of the radio range of source node, other nodes are used as relay stations [6].
Geographic Routing (GR) is one of the schemes which have gained most momentum in the recent years. In GR, each node needs to know the position of its neighbors. To do that, they send periodic HELLO messages called beacons including the identifier of the sender and its position. These packets are not forwarded; therefore only one hop neighbors can receive them. The packets include the identifier of the sender and its position.
However, although geographic routing in general is
highly desirable, the beaconing mechanism has
some issues, such as generating interferences with
regular data transmission and consuming
bandwidth and battery power. In particular in those
sensors not taking part in any routing process, the
energy and bandwidth consumption represent a
total waste of resources. To overcome such issues,
a new routing protocols class that exploits the
cross-layer interactions was proposed, i.e.,
receiver-based forwarding. Receiver-based routing
protocols are highly scalable and robust against
frequent topological changes [
In [7] and [8], the performance analysis of the
geographical random forwarding (GeRaF)
algorithm is presented. This algorithm introduces
receiver-based routing for cross-layer interaction
between MAC (Medium Access Control) and
routing layers. However, the GeRaF algorithm requires
a sensor node with two radios for signaling, which
may not be feasible in some scenarios. In [
A cross-layer module (XLM) proposed in [
Receiver-based contention techniques have also
been adopted in several cross-layer MAC and
routing protocols. The adaptive load balanced
algorithm (ALBA) described in [2], [3] and [4], is
based on the GeRaF framework in [7]. In addition
to the location of the nodes, the traffic load on each
node is considered for route establishment. More
specifically, each potential node computes two
values: geographic priority index (GPI) and queue
priority index (QPI), which indicate the progress of
the node towards the destination and its traffic load,
respectively. Accordingly, if a node has a packet to
send, it sends several RTS packets to scan QPI
and GPI values of its neighbors. Each neighbor
responds to this packet by CTS packets if their
values match the requested values. The source
node then selects one of the neighbors if the
requested value is found. Moreover, ALBA-R
proposed in [4], enhances the original protocol to
avoid local minima in routing through a coloring
scheme called Rainbow. Based on its previous
success in finding relays, each node assigns itself
a different color, which is used to participate in
communication. While ALBA-R employs a
crosslayer MAC/routing technique, the route selection is
performed at the sender node, which incurs high
overhead due to the QPI and GPI scanning.
In [
In [5], authors have proposed RSSI-Based Forwarding (RBF), a cross-layer integrated medium access control/routing protocol for multi-hop WSN. Without using prior knowledge of nodes’ geographical locations and without maintaining neighborhood routing tables, the next-hop node for data forwarding task is determined at the same time as the contention process among the possible forwarding nodes is solved. For an arriving beacon signal transmitted by the sink, received power levels are computed for each sensor node in the network and these levels are then used as the decision parameter for the nodes to contend for the forwarding task of the data packets. RBF protocol is very simple; the localization of the nodes is done by a simple broadcast by the sink. Nevertheless, the nodes taking part in the contention process have all the same probability of being selected for forwarding the packet, which generates a significant number of hops in the routing process. Moreover, the authors proposed to increase the power of transmission to overcome the problem of holes. This solution is really not possible in sensor networks, because the increase in transmission power causes more collisions and consumes more energy.
Most of the receiver-based routing protocols
reported in the literature assume knowledge of
nodes’ geographical locations for routing the
sensed data toward the sink node. With the WSN’s
characteristics that require a large number of
lowcost and energy-efficient sensor nodes, equipping
a GPS on every sensor node may not be practical.
Furthermore, the cost of a GPS chip is much more
expensive than the sensor node itself [
CLEAP protocol runs in two phases: the network setup phase and the data communication phase.
The sink initiates the connection by broadcasting a Beacon frame. Transmission power of the sink must be large enough so that the Beacon signal reaches all nodes in the network. Each node in the network stores the Received Signal Strength Indicator (RSSI) in a variable my_RSSI to use as a routing parameter.
The main role of this phase is to transmit data collected by the sensor nodes to the sink. It consists in choosing the next node that will receive the data to transmit in its turn to others. This process is repeated until data reach its destination. This phase is executed in two steps: one for the transmission initiation and the other for the contention process.
This step is based on the CSMA/CA mechanism of IEEE 802.11 [11]. It is triggered when a node has data to transmit. In this case, it puts his value my_RSSI in a field RSSI of RTS (Request To Send) packet and performs the CSMA/CA algorithm to broadcast it. Then, the node starts a timer CTS_Wait to wait for a CTS (Clear To Send) response. If CTS_Wait expires without having received a CTS response, the node performs Backoff and retransmits an RTS. For each RTS retransmission, a counter of retransmissions number, Short_Retry_Count, is incremented. In this case, the RTS retransmission will take place if this counter is below a predefined threshold Short_Retry_Limit. When a CTS is received, the counter is reset to zero.
Each node receiving the RTS packet determines which region it is as follows: if the value my_RSSI is strictly greater than that included in the RTS packet, the node is considered in the relay region F, otherwise it is considered in Fc. Nodes that are within the relay region will participate in the contention process, while other nodes will go to sleep for the duration of NAV (Network Allocation Vector) specified in RTS packet.
This step is based on the routing priority of each node; the priority is determined by two parameters: the packet progress and residual energy. To do this, each node calculates its optimality index Iop, determines its priority class corresponding my_classe and starts a timer CTS_Response to wait before responding by CTS.
Where Er is the residual energy of node and RSSI is the value my_RSSI of the node sending the RTS (that is included in the RTS). RSSI_max is the maximum value of the parameter my_RSSI that can have a relay node; it represents the RSSI value at a distance of (D-r) of the sink, where D is the distance between the source node of RTS packet and the sink. r is the radius of transmission. To reduce the likelihood that multiple nodes simultaneously send their CTS, the relay region F is divided into Nc1 priority classes as shown in Figure 1. A node determines its class by the following equation: {⌊ ⌋ Where last_Iop is a parameter included in the RTS (set to 1 by default), it represents an estimate of the largest value Iop of nodes belonging to the relay region. In this way, we obtain a relay region divided into Nc priority classes. Class 0 contains the most optimal nodes towards the metric Iop and class Nc-1 contains the nodes less optimal. Once the priority class my_classe is determined, the node starts a timer CTS_Response before responding by a CTS.
( ∑ ) ( ) Random (1, cw) is a function that generate a timeslot between 1 and cw (Contention Window). It’s obvious that the nodes of class Ni have a waiting time strictly less than the waiting time for node class Ni+1. In addition, the nodes of the same class may have different waiting times thanks to the function Random ( ).
After CTS_Response expire, the node responds with CTS by setting its value Iop in the field last_Iop of the CTS. To avoid multiple responses CTS, each node listens to the channel during his waiting time CTS_Response. If a CTS is heard, the node concludes that another node has sent his CTS, in this case it cancels its timer, updates its NAV and goes to sleep. If a data packet is detected, the node cancels its timer, updates its NAV and goes to sleep. In the case where two nodes send their CTS simultaneously, the Backoff mechanism is used to resolve the collision problem. After receiving the CTS packet from the initiator node, it updates its value last_Iop, transmits the data packet and starts a timer ACK_Wait = SIFS to wait for an acknowledgment (ACK). If this timer expires 1 This number has been tuned via extensive simulation and selected to obtain better performances in terms of energy consumption and latency. without receiving an
ACK, the source node assumes a packet loss and retransmits the data packet again. For each DATA retransmission, a counter
Long_Retry_Count of retransmissions number is incremented. DATA retransmission will take place if this counter is below a predefined threshold Long_Retry_Limit, otherwise the packet will be dropped. After receiving the ACK packet, the counter is reset. The amount of time set for the CTS_Wait timer corresponds to the maximum time allowed to
send
In Figure 1, node 1 sends an RTS packet and initiates a contention process in its neighborhood. Nodes in the relay region determine their priority classes as shown in the figure. Node 15 waits less time than other nodes to respond by CTS, as shown in Figure 2, nodes 9, 8, 5, 45 and 7 hear the CTS of node 15 and goes to sleep. After receiving the
CTS packet by node 1, an exchange
0 2 3 −
Relay Region F Region Fc 45 7 8 9 Sink 3 ...
CTS CTS CTS
response, i.e., discarding the packet. [4]. A node which cannot advance packets toward the sink switches its color to ‘red’. Red nodes handle a packet that they generate or that they receive according to a different rule: the packet is sent away from the sink selecting as relay green or red nodes in Fc. This process is repeated until a green node is reached. Starting from the green node, CLEAP operations’ are resumed. The packet is forwarded to the sink along a route which goes only through green nodes. If a red node is unable to find relays in Fc, it progressively stops proposing itself as relay for other red nodes, eventually
In this section, we illustrate the performance
evaluation of CLEAP. The existing sensor network
simulation platforms are not suitable for cross-layer
communication suite design due to their layered
architecture
[
CLEAP and RBF in our simulator developed at our laboratory in
C#.
Our simulator consists of a channel model and an event-driven simulation engine. The simulations are done in random topologies with different sets of 50 to 130 nodes. Each simulation lasts for 100 seconds, and the results are the average of 30 trials for each of five different random topologies. Sensors are deployed in a square area with side L=150 m.
1 show some parameters used in the simulation. To not penalize RBF, we have chosen, as recovering strategy, to increase the transmission power at 14 mW when meeting a hole. in [10]. Equation (4) gives the path loss PL at a distance d from the transmitter node where PL(d0) is the path loss at a reference distance d0, η is the path loss exponent.
0
According to this model, the RSSI_max, explained
in section three for Iop calculation, is obtained by:
Where Pt is the transmission power of the Beacon
signal transmitted by the sink. Note that when the
forwarding node is searched among nodes in relay
region F, these nodes use the sign (-) in the
equation, otherwise they use the sign (+).
The energy consumed by the nodes is calculated
through the first rder energy model described in
[
In Figure 3 (a) and (c), we present the average residual energy (ARE). Results in (a) shows that CLEAP improves average residual energy in the network compared to RBF protocol. This improvement can be justified by the mechanism used by CLEAP protocol that tries to route the packet through the optimal path in terms of energy consumption. For cons, RBF protocol routes packets through a path chosen randomly, which can be costly in terms of energy consumption and especially when there is a hole. However, when the network is less dense (70 nodes), RBF has better performances than those obtained with CLEAP. This energy saving is obtained through the recovering mechanism applied by RBF, which is to increase the transmission power for out of a hole. In fact, this mechanism is not effective in denser networks. From Figure 3 (c), we can see with RBF, nodes deplete their energy quickly compared to our protocol CLEAP. And this proves the effectiveness of our mechanism applied during the timers’ calculation, taking into account the residual energy of nodes participating in the contention process. Figure 3 (b) shows the average delivery rate (ADR) results. From the figure, we can observe that the delivery rate is proportional to the number of nodes in the network. It can also be noted that when the number of nodes is less than or equal to 70 nodes, the average delivery rate of RBF protocol is better than that of our protocol CLEAP because of the number of nodes disconnected (note that we did not impose any constraint on the connectivity of the network when generating topologies). However, beyond 70 nodes, our protocol gives better delivery rate than the RBF protocol.
We note from Figure 3 (d) that when the number of nodes is less than 60 nodes, the average latency (AL) obtained with the RBF protocol is better than that obtained with our protocol, and this is reflected by the low density of the network. In fact, when the network is less dense, increasing the transmission power is better suited than the perimeter routing. However, when the number of nodes is greater than 60, the average obtained with our protocol is better than that obtained with RBF. The reason is that our selection mechanism takes into account the packet progress which minimizes the number of hops when routing data packets.
Conceiving energy-efficient protocols is a critical issue in energy-constrained wireless sensor networks. In this paper, we proposed an integrated cross-layer MAC/routing protocol; named CLEAP. By exploiting interaction with the physical layer, our proposed scheme does not need a localization protocol which eliminate the costly techniques to determine the sensor nodes’ locations; the estimate of the distance which separates each node from the sink, is done by broadcasting a beacon signal from the sink, the measured RSSI is then combined with the remaining energy in the contention process. Energy consumption and latency has been considered as the important indicators to analyze the effectiveness of the protocol in eventdriven WSN, and the protocol has been shown to have low latency and low energy consumption as well as robust data delivery rate.
Much further research remains to be done on how to add local parameters, such as interferences level and buffer occupancy, to the CTS_Response without altering basic system behavior. Furthermore, we want to study the error measurement impact of the RSSI on the performances of the protocol. 0,4 )s 0,3 e l u j(o 0,2 E R A0,1
0 100 80 ) 60 % ( R AD40 20 0
RBF CLEAP
RBF
CLEAP 130 150 50 70
90 110 130 150 number of nodes [3] P, Casari. M, Nati. C, Petrioli. M, Zorzi. (2007) Geographic Forwarding and Adaptive Load Balancing in Wireless Sensor Networks. ACM SIGMOBILE Mobile Computing and Communications Review, vol. 11(2), 2007, 5354. [4] P, Casari. M, Nati. C, Petrioli. M, Zorzi. (2007) Efficient Non Planar Routing around Dead Ends in Sparse Topologies using Random Forwarding. IEEE International Conference on Communications, ICC 2007, 3122-3129. [5] A, Awang. X, Lagrange. D, Ros. (2009) A Cross-Layer Medium Access Control and Routing Protocol for Wireless Sensor Net-works. 10èmes Journées Doctorales en Informatique et Réseaux, JDIR 2009, 85-90. [6] J.A, Sanchez. R, Marin-Perez. P.M, Ruiz. (2007) BOSS: Beacon-less On Demand Strategy for Geographic Routing in Wireless Sensor Networks. IEEE Conference on Mobile Ad-hoc and Sensor Systems, MASS 2007, 1-10. [7] M, Zorzi. R.R., Rao, “Geographic Random Forwarding (GeR-aF) for Ad Hoc and Sensor Networks: Multihop Performance,” IEEE Transaction on Mobile Computing 2, 2003, 337348. [8] M, Zorzi. R.R, Rao. (2003) Geographic Random Forwarding (GeRaF) for Ad Hoc and Sensor Networks: energy and latency performance. IEEE Transaction on Mobile Computing 2, 2003 349-365. [9] P, Skraba. H, Aghajan. A, Bahai. (2004) Crosslayer optimization for high density sensor networks: Distributed passive routing Decisions. International Conference on Ad Hoc Networks and Wireless, AD-HOC NOW 2004,266-279. [10] T.S, Rappaport. (2001) Wireless Communications: Principles and Practice. Prentice Hall, New Jersey. [11] IEEE Standard for Information TechnologyTelecommunications and Information exchange between systems-Local and metropolitan area networks-Specific requirements. Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) specifications: Higher-Speed Physical Layer Extension in the 2.4 GHz Band. (2003)