Hidden and Exposed Node Terminal Problem

Due to the limited spectrum availability and the ever increasing number of wireless devices, the design of the MAC (media access control) layer is therefore has to be efficient in reducing channel contentions amongst users. An improper channel assignment could lead to an increase in the global interference which would increase delay and hence the energy consumption of the network as nodes will be in the transmitting and receiving mode for a longer duration. The high interference will also reduce the energy efficiency as nodes will be required to retransmit messages caused by dropping.

The topology of a wireless network is such that transmitting devices may or may not interfere with its neighbours which give rise to the two well know problems of hidden-node and exposed node problems [85], [87]. Figure 5-5 illustrates such problems. The hidden node terminal problem occurs when both node a and node c simultaneously transmit on the same channel to node b which causes the packet to collide resulting in performance degradation. Employing the ability to sense before transmission such as CSMA (Carrier Sense Multiple Access) does not mitigate the hidden node terminal problem as in the case in Figure 5-5, node a is outside the transmission range of node c and is therefore cannot sense the existence of each other.

Consider that there is an ongoing transmission from node b to node a. as shown in Figure 5-5b and assume that the nodes access the channel via CSMA/CA (Carrier Sense Multiple Access with collision avoidance protocol). If node c wants to transmit to node d, it will first sense its environment and inaccurately deduced that there is an ongoing transmission that will cause a collision if it transmits messages onto the same channel. Such transmissions however, would only cause a slight increase in interference at node a. This unnecessary delay in transmission would result in the channel being underutilised and thus reduce spatial reuse; such a problem is known as the exposed node terminal problem.

a b c

a b c d

(a) Hidden terminals

(b) Exposed terminals

Figure 5-5: Hidden and Exposed node problems, each circle indicates the transmission range of the node at its centre

In IEEE 802.11 [85], the means in which it accesses a channel is called the distributed coordination function (DCF).DCF partially solved the issue with hidden node terminal problem in the scenario as shown in Figure 5-5 with the introduction of a hand shaking technique onto CSMA/CA [86]. The hand shaking technique requires the sender to ‘test’ the state of a channel by sending a short frame of RTS (Request to Send) to the receiver. If the RTS is successfully received by the receiver, it will respond by sending CTS (clear to send) frame to the transmitter. In the event of the channel being occupied by another transmission, the RTS frame will not be successfully received by the receiver due to collision and there will be an absence of a CTS response. The handshaking protocol avoids collision of packets and thus increases the network performance. It is worth noting that the hidden node terminal problem still exists in a multi-hop network [87] and [88]. However, the implementation of the handshaking protocol still does not solve the exposed node terminal problem [87], [89] and [90].

In a dual hop clustered network as shown in Figure 5-6, the hand shaking protocol does not mitigate the existence of hidden node terminal problem. Consider that there is an uplink communication from cluster member bm to cluster head bh and that the transmit power level of node bm is limited such that only its respective cluster head bh can successfully receive the transmission. The ongoing uplink communication from node bm to cluster head bh will be ‘hidden’ from cluster head ah and thus it will not be able to sense the ongoing communication. If there is a message to be transmitted via the backhaul link from cluster head ah to hub base station (HBS), it will falsely conclude that the channel is empty and began transmission. Due to the nature of the dual hop clustered architecture, some clusters are located at the edge of the network and the maximum distance of cluster head transmission range (m) assuming a squared network with l side lengths is:

√ 5.11

The maximum transmission range of cluster heads drmx will affect hidden uplink

transmissions within an area of:

5.12

From (5.12), it can be seen that the maximum transmission range of cluster heads can cover an area of almost 40% of the network (assuming squared network area). Due to the cumulative transmission range of other cluster heads in the network, substantially more nodes are in the vulnerable region i.e. a node transmitting to its respective cluster head cannot be detected by neighbouring cluster heads and can be disrupted due a backhaul connection from neighbouring clusters occupying the same channel.

The cumulative effect of the hidden node terminal problem can severely affect the uplink transmission thus reducing the scalability of the dual hop clustered network. To mitigate the hidden terminal problem posed by the large interference range of the cluster head transmission, a split channel pool bandwidth (assuming the total number

of channel available in the network >2) between the uplink and backhaul connections is proposed. The separate channels for the uplink allow the communication to be uninterrupted by the backhaul connections from cluster heads in the network. Splitting the common channel pool however limits the number of available backhaul connections which could compromise the capacity of the network.

Uplink Backhaul connection

Up link HBS am bm ah bh Vulnerable region

Figure 5-6: Shaded area indicates the vulnerable region in which uplink transmissions cannot be detected by cluster head ah