Guaranteeing Quality of Service in Mobile Ad Hoc Networks

Defining Qos

QoS is a network performance concept whose definition varies depending on the treatment given to the traffic belonging to a determined user’s profile (Juliet Bates, 2006). For example, video applications require sufficient bandwidth and low packet loss. VoIP services need low latency, but can withstand a limited packet loss.

From a network perspective, QoS is characterized by a number of quantifiable attributes. The following are examples of the QoS commitments that a packet network can give:

• •

  Within a call, the packet loss, end-to-end delay and latency are the main parameters. Such parameters can be explicitly specified in the service level agreement, giving rise to “hard” QoS commitments. Alternatively, “soft” QoS commitments may be given that imply priority of a user’s traffic.

• •

  The network maintains a model of resources management and decides, with respect to user’s profiles, if it can admit traffic while achieving QoS. This is known as admission control.

For these mechanisms to achieve a quantifiable level of QoS, the profile of each user must be understood. In mobile networks, MANETs should know how to manage resources and meet the QoS commitments. The challenge in these networks consists in delivering the appropriate levels of QoS while ensuring that they can scale to all users and balancing this with the cost and risk of over-provisioning.

Admission Control Enabled On-Demand Routing Protocol

Designing an efficient and reliable QoS routing protocol for MANETs is a challenging problem (X. Masip-Bruin, 2006; S.R. DAS, 2000). However, a simple routing mechanism is required to efficiently manage the limited resources while at the same time being adaptable to the changing network conditions.

ACOR (N. Kettaf H. A., 2006) was proposed to efficiently provide end-to-end support for QoS by introducing simple cost functions which represent QoS metrics. Inspired by the colored subgraphs formulation presented in K.M. Konwar (2005), where each link of a MANET is divided potentially to sublinks represented by elementary cost functions for QoS metrics (i.e., bandwidth, delay, packet loss, etc.). For purposes of clarity, we only focus on bandwidth and delay. Firstly, the bandwidth at each node is represented by Fb, which is a ratio of the requested bandwidth B by an application to the supported bandwidth by a link Bmax in addition to the residual bandwidth Bres. Secondly, the delay is represented by Fd which is also a ratio of supported delay D by an application to the accumulated hop-by-hop delays with the upper bound of delay Dmax. The sum of the elementary cost functions (Fb and Fd) at each node is added to the global cost function Fg received in the route request packet during the route discovery to represent a route’s end-to-end cost.