Contention Based Protocols
Contention Based Protocols
Contention Based Protocols
1. Introduction
Ad hoc wireless networks are comprised of mobile nodes that exchange packets by sharing a common broadcast radio channel. Due to the limitations of this channel, the bandwidth to be shared among the nodes is limited. Therefore, the aim in these networks is to be able to utilize the bandwidth efficiently, and guarantee fairness to all nodes. As we know, wireless networks differ enormously from wired networks; furthermore, ad hoc wireless networks have even more specific characteristics, such as node mobility, power constraints. Thus, new protocols are needed for controlling access to the physical medium. The unique properties of the ad hoc networks make the design of a media access control (MAC) protocol more challenging. This paper sets on giving a brief outline of the MAC protocols for ad hoc networks, focusing on contention-based algorithms with reservation and scheduling.
2. Design issues
Following are the main issues one should have in mind when considering designing a MAC protocol for ad hoc wireless networks. Bandwidth efficiency: The scarcity of bandwidth resources in these networks calls for its efficient usage. To quantify this, we could say that bandwidth efficiency is the ratio of the bandwidth utilized for data transmission to the total available bandwidth. In these terms, the target will be to maximize this value. Quality of service support: Providing QoS in these networks is very difficult, due to the high mobility of the nodes comprising them. Once a node moves out of another nodes reach, the reservation in it is lost. On the other hand, in these networks QoS is sometimes extremely important, for example in military environments. Therefore, QoS should be provided somehow, despite the characteristics of ad hoc networks. Synchronization: Some mechanism has to be found in order to provide synchronization among the nodes. Synchronization is important for regulating the bandwidth reservation. Hidden and exposed terminal problems: The reason for these two problems is the broadcast nature of the radio channel, namely, all the nodes within a nodes transmission range receive its transmission. Hidden terminal problem two nodes that are outside each-others range perform simultaneous transmission to a node that is within the range of each of them, hence, there is a packet collision.
Page 1
Figure 1: Example of the hidden terminal problem Exposed terminal problem the node is within the range of a node that is transmitting, and it cannot transmit to any node.
Figure 2: Example of the exposed terminal problem Hidden nodes mean increased probability of collision at a receiver, whereas exposed nodes may be denied channel access unnecessarily, which means underutilization of the bandwidth resources. Error-prone shared broadcast channel: In radio transmission, a node can listen to all traffic within its range. Therefore, when there is communication going on no other node should transmit, otherwise there would be interferences. Access to the physical medium should be granted only if there is no session going on. Nodes will often compete for the channel at the same time; therefore, there is high probability of collisions. The aim of a MAC protocol will be to minimize them, while maintaining fairness. No central coordination: in ad hoc networks, there is no central point of coordination due to the mobility of the nodes. Therefore, the control of the access to the channel must be distributed among them. In order for this to be coordinated, the nodes must exchange information. It is the responsibility of the MAC protocol to make sure this overhead is not a burden for the scarce bandwidth. Mobility of nodes: The mobility of the nodes is one of its key features. The QoS reservations or the exchanged information might become useless, due to node mobility. The MAC protocol must be such that mobility has as little influence as possible on the performance of the whole network. Signal propagation delay: Signal propagation delay is the amount of time needed for the transmission to reach the receiver. If the value of this parameter is considerable, a node may start transmitting, when in fact, transmission from other nodes is taking place, but it has not reached the node yet. The ad hoc networks that utilize synchronization, therefore, will have to expand the time slot to accommodate the propagation delay. Hardware constraints: Most radio-receivers are designed in such a way that only halfduplex communication can take place. When a node is transmitting, the power level of the outgoing signal is higher than any received signal; therefore, the node receives its own
Page 2
transmission. Here, we can also add hardware switching time time needed to shift from one mode to the other.
Page 3
Figure 3: Classification of the MAC protocols for ad hoc networks General definition of contention-based protocols Here, the channel access policy is based on competition. Whenever a node needs to send a packet, it tries to get access to the channel. These protocols cannot provide QoS, since access to the network cannot be guaranteed beforehand. General definition of contention-based protocols with reservation mechanisms These protocols provide bandwidth reservation ahead; therefore, they can provide QoS support. These can be further subdivided into: - Synchronous protocols: there is time synchronization among all nodes in the network, the nodes in the neighborhood are informed of the reservations; - Asynchronous protocols: no global synchronization is needed. Relative time is used for the reservations. General definition of contention-based protocols with scheduling mechanisms There can be packet scheduling at the nodes, or node scheduling for access to the channel. Node scheduling should not treat the nodes unfairly. Some of these protocols consider battery power in their node scheduling.
Page 4
Figure 4: Time Division in the D-PRMA protocol The mechanism of competition for slots is such that a certain period at the beginning of every slot is reserved for carrier-sensing. The nodes compete for the first minislot in each slot. The winning one transmits a RTS packet through the RTS/BI part of the first minislot. The receiver responds by sending a CTS packet through the CTS/BI field. Thus, the node is granted all the subsequent minislots [1]. In addition to that, this very same slot in the subsequent frames is reserved for the same node, until it ends its transmission. Within a time slot, communication between the source and destination nodes is done either by Time Division Duplexing (TDD), or by Frequency Division Duplexing (FDD). There are two rules for the reservation, which prioritize voice traffic: - Contention for the first minislot is done with probability 1 for voice traffic, and a smaller probability for other traffic. - The reservation of a minislot brings reservation of the subsequent slots only if the winning node is a voice one.
Figure 5: Time Division in the CATA protocol CATA supports broadcast, unicast, and multicast transmissions at the same time. CATA has two basic principles: - The receiver of a flow must inform other potential source nodes about the reservation of the slot, and also inform them about interferences in the slot. - Negative acknowledgements are used at the beginning of each slot for distributing slot reservation information to senders of broadcast or multicast sessions. The CMS1 and CMS2 are used to inform neighbors of the receiving and the sending nodes accordingly about the reservation. The CMS3 and CMS4 are used for channel reservation [1].
Page 5
Figure 6: Time Division in the HRMA protocol One frequency channel is a dedicated synchronizing channel where nodes exchange information. The remaining frequency channels are paired, one channel in each pair is used for reservation and data packets, and the other one is used for acknowledgements. As mentioned above, each time slot has a frequency channel. The time slot is divided into four periods, each period is reserved for sending a particular kind of packet or its acknowledgement, depending on which frequency channel of the pair this time slot belongs to. After the handshaking is over, the two nodes communicate by sending data and ACKs on the very same frequency channels. When a new node wants to join the network, it listens to the dedicated frequency and gathers information. When a node wants to send data, it listens to the Hop Reservation (HR) period. If there is a packet there, it tries again after a random amount of time, otherwise it sends a RTS packet, and waits for the CTS acknowledgement packet in the CTS period of the corresponding frequency channel.
Page 6
Figure 7: Packet exchange in MACA/PR For real-time traffic, the first part is identical until the first data packet is sent. Each data packet contains information about the reservation of the next data packet. This information is piggy-backed to it. Each acknowledging packet also contains this information. Thus, the neighbors of both communicating nodes can update the information. When the sender receives the acknowledgement, it makes sure that the reservation was successful. If several acknowledgements do not come, the sender assumes that the reservation has been lost and restarts the whole procedure. The acknowledgement refreshes the reservation; unrefreshed ones are simply dropped from the reservation table. The nodes exchange the information in their reservation tables; this eliminates the hidden terminal problem. This mechanism works as a TDM for real-time traffic, while best-effort packets are transmitted in the empty slots. When a new node joins the network, at first it learns about it by receiving the reservation tables from the neighbors. Then it behaves just like the others. Advantage: global synchronization not required. Drawback: the RTS-CTS-DATA-ACK exchange takes place in the same slot in different cycles; therefore, random empty slots are not utilized.
Page 7
7. Conclusions
The issues associated with the design of a MAC protocol for wireless ad hoc networks are: node mobility; an error- prone, broadcast and shared channel; time-synchronization; bandwidth efficiency; QoS support.
8. References
[1] C. Siva Ram Murthy and B. S. Manoj: "Ad Hoc Wireless Networks: Architectures and Protocols" [2] V. Kanodia, C. Li, A. Sabharwal, B. Sadeghi , and E. Knightly: Ordered Packet Scheduling in Wireless Ad Hoc Networks: Mechanisms and Performance Analysis [3] Alexander Tyrrell, Gunther Auer and Christian Bettstetter: Firefly Synchronization in Ad Hoc Networks
Page 8