The Carrier Sensing Multiple Access (CSMA) protocol requires a node to listen to the shared channel before commencing a transmission (called the carrier sensing) to check if the channel is idle or occupied by a transmission [13,22]. If the channel is found idle then a node starts transmission otherwise a node defers its transmission (i.e., backs-off) for a random amount of time. Based on the policy by which a node assesses the channel and the back-off scheme, a number of variations of CSMA were proposed e.g., non-persistent, 1-persistent and 𝑝- persistent protocols.
In the non-persistent CSMA scheme, whenever a node assesses the channel as busy, it backs-off for a random amount of time before assessing the channel again. A node does not care about the state of the channel during the back-off intervals. The major drawback of the non-persistent protocol is that it may lead to underutilisation of the channel capacity as the channel may become idle while a node is backing-off.
In order to resolve this drawback, the 1-persistent CSMA protocol has been proposed. In the 1-persistent protocol, a node whenever assessing the channel as idle commences transmission immediately; otherwise a node assesses the channel persistently until the channel becomes idle. Hence the 1-persistent CSMA protocol can increase the channel utilisation as a node commences its transmission as soon as the channel becomes idle. However, the main drawback of the 1-persistent is that it consumes a considerable amount of energy in ceaseless sensing.
The 𝑝-persistent CSMA scheme was introduced to take advantage of both non- persistent and 1- persistent schemes. In the 𝑝-persistent CSMA scheme, a node that finds an idle channel commences transmission with probability 𝑝 and with the complement of this probability a node waits for a predefined period and then attempts to transmit a packet again. This waiting period is set either to the propagation delay between the most distant nodes in the unslotted CSMA scheme or to a slot when the slotted CSMA scheme is used. When the waiting period has elapsed, a node senses the channel again; if it is found busy, a node senses the channel persistently until it becomes idle and then repeats the aforementioned process again. The main limitation of the 𝑝-persistent CSMA scheme is that it requires an accurate adjustment of the value of probability 𝑝 with respect to the traffic intensity. Adjustment of 𝑝 to a large value under high traffic intensity can increase the collision probability of packets and lessen the throughput of network. The main reason for this characteristic is that a large value of 𝑝 allows nodes to commence their transmissions almost immediately, which in turn increases the contention to access the channel and magnifies the number of colliding nodes. On the other hand, setting the value of the probability 𝑝 to a small value under light traffic rates can lead to considerable underutilisation of the channel, as a node potentially waits for longer periods while the channel is idle. The main challenge of employing the 𝑝-persistent CSMA scheme to multihop WSNs is that the traffic characteristics of these networks are highly dynamic and differ substantially amongst nodes. This in turn hinders prediction the appropriate value of 𝑝 that suits these traffic conditions.
M. Baz, PhD Thesis, University of York 2014
Comparing the performance of the ALOHA protocols with CSMA protocols demonstrates that the CSMA protocols are able to improve the throughput of network. CSMA enables a node to listen to the shared channel before commencing its transmissions which in turn eliminates the possibility of transmissions while the channel is occupied. However, the CSMA scheme suffers from high packet delay, as a node has to wait until the reception of the acknowledgment packet from the receiver to decide whether the transmitted packets have been received successfully or not.
With the aim of minimising the delay of the CSMA protocols, the Carrier Sensing Multiple Access with Collision Detection (CSMA-CD) protocol was proposed. CSMA-CD like the aforementioned CSMA protocols requires a node to assess the channel before transmission. If a node finds an idle channel then it transmits its packet immediately. Thereafter a node continues to monitor the transmitted signal and stops its transmission as soon as a collision is detected. Colliding nodes back off for a random duration generated according to a truncated binary exponential back-off algorithm based on the length of slot and the number of retransmissions. The main advantage of CSMA-CD is that it mitigates the delay of packets and increases the throughput of network by minimising the loss of channel capacity. However, this scheme cannot be used over most wireless networks, as detection of the occurrence or absence of collisions by the transmitter over the wireless channels is difficult due to the fluctuation of wireless channels and the decreasing of power of the signal over distance and time. Hence, a new contention based multiple access scheme called Carrier Sensing Multiple Access-Collision Avoidance (CSMA-CA) was proposed [13-32].
CSMA-CA leverages the performance of the CSMA protocols by reducing the possibility of collisions instead of detecting them as is the case in the CSMA-CD. The CSMA-CA protocol endeavours to solve two common problems found in wireless communications; namely, the hidden node and exposed node problems [13, 32]. A hidden node is a node that is within the range of the receiver but out of the range of the transmitter, conversely an exposed node is a node that is within the range of the transmitter but is out of the range of the receiver. Illustrations for the definition of the hidden and expensed nodes are given in figure 3.3 (a) and (b) respectively.
M. Baz, PhD Thesis, University of York 2014 Figure 3.3 Hidden and exposed node scenarios
Figure 3.3(a) depicts a hidden node scenario; it shows that node B is within the transmission ranges of both nodes A and C but node A and C are outside their mutual transmission ranges. Hence a transmission from either A or C can be heard by B but a transmission from A cannot be received by node C and vice versa. In order to illustrate the scenario for hidden node collisions, let us assume that node A intends to transmit a packet to node B. Node A assesses the channel, finds it idle and commences its transmission. Assuming that during transmission of node A towards B, node C wishes to transmit a packet to node B. Here node C assesses the channel and finds it idle (as the transmission from node A cannot be heard by node C). As a result C commences its transmission to node B. Therefore, both transmissions from A and C will collide at node B. This collision is known as a hidden node collision. Figure 3.3 (b) depicts the exposed node problem, it shows that node B is within the transmission range of both nodes A and C. Nodes A and C are outside their mutual transmission ranges and node D is within the transmission range of node C. Assuming that node B needs to send a packet to node A, node B assesses the channel, finds it idle and commences the transmission. During this transmission, node C wants to send a packet to node D. When node C assesses the channel, it finds a busy channel, as there is an ongoing transmission from node B to node A. So node C defers its transmission, however, it can be seen from figure 3.3 (b) that the transmission from node C to D and transmission from B to A can be carried out simultaneously and delivered successfully, as node D is outside the transmission range of node B.
In order to alleviate the hidden and exposed problems, several techniques have been proposed. The first technique is called the busy tone [13]. This technique is inspired by the fact that the collision takes place at the receiver while CSMA is performed at the transmitter nodes. Busy tone uses two distinct radio transceivers, one for data and the other for the control. The data channel is only used to transmit data packets while the control channel is used by the receiver to notify its neighbours that it is busy receiving a packet. Hence when a node wishes to transmit a packet, it first senses the carrier of the control channel. If there is busy tone signal then the node defers its
A B C Transmission range of node A Transmission range of node B Transmission range of node C (b) A B C D Transmission range of node A Transmission range of node B Transmission range of node C Transmission range of node D (a)
M. Baz, PhD Thesis, University of York 2014
transmission. The busy tone can solve both of the hidden and exposed node problems, however the need to equip each node with a two transceivers or a full duplex transceiver increases the complexity of tiny nodes which in turn pushes their production cost up and more importantly leads to fast depletion of their batteries. The second technique that has been proposed to mitigate the effects of hidden node collision is dubbed Ready To Send (RTS) and Clear To Send (RTS) mechanism [79]. In such an approach, a transmitter sends the RTS packet to the intended receiver after assessment the channel as idle. The RTS packet is addressed to the intended receiver and contains information about the length of data packet. When the intended receiver receives the RTS packet, it assesses the channel and sends the CTS packet. If the transmitter receives a CTS packet it commences transmission of its data packet, otherwise the transmitter waits for a random time and then repeats the process again. Although the RTS/CTS mechanism can reduce the probability of hidden node collisions in some scenarios, the overheads associated with exchanging RTS/CTS packets as well as their energy costs make this mechanism unsuitable for the limited resource WSNs.