User-input driven QoS management in ad hoc networks

Abstract

Though the network quality-of-service (QoS) metrics are defined in terms of technical parameters (e.g., delay, jitter, bandwidth), they are rather subjective when it comes to the end user. Oftentimes, the end user finds it difficult to express his desired QoS in such technical parameters, though he has a fair idea of what QoS he desires. In this paper, we show how translation functions can be devised and used to translate the user inputs to networking parameters that are used by various layers of the protocol stack. In particular, we consider an ad hoc network and show an interface design that uses translation functions to map user supplied inputs to parameters at the medium access control (MAC) and routing layers. These parameters, in turn, choose the right strategies that particular layer functionality can adopt, such that the QoS desired by the user is achieved. We also compute the associated costs due to the different strategies adopted. We implement the interface on ns-2 and conduct simulation experiments with randomly scattered IEEE 802.11 enabled nodes. Results show the functionality of the interface and demonstrate how delay, throughput, and network lifetime are affected when the end user seeks different levels of QoS.

Introduction

Ad hoc networks play a critical role in places where a wired (central) backbone is neither available nor economical to build, such as law enforcement operations, battle field communications, disaster recovery situations, and so on. Such networks comprise a multitude to devices that can communicate with each other directly as long as the transmitter and receiver are within the transmission range of each other. The absence of a central entity, like an access point or a base station, makes it difficult to enforce any kind of coordination among the hosts (also called as nodes). This lack of coordination severely hampers the packet transmission activities of the nodes and hence the end-to-end performance of the system.

In situations where user specified quality-of-service (QoS) is desired, a straightforward and easily manageable QoS interface is required that can translate what the user desires to what the network ought to do. The QoS provided by a network typically refers to the performance yield of the lower layers of the protocol stack such as network, data link, and physical layers. However, end-to-end issues also involve performance optimizations at the higher protocol layers. Though not being part of the protocol stack, the end users using the network and deriving services can also play a critical role in deciding and manipulating the desired level of QoS [16]. It must be noted that for every network transaction that a user makes, there is usually a cost associated that the user/network must agree to pay. This cost (e.g., energy consumption, compromised QoS) is merely a mechanism that allows one user to get better service than another.

One of the most important performance metrics of a network is the end-to-end delay. As far as controlling the delay is concerned, it is important that the network operates such that the performance is within a specified delay budget. Due to the distributive nature of ad hoc network, adhering to such hard delay guarantees is not possible. However, efforts can be made by the nodes that try to maximize the probability of yielding a certain delay. Any application or user imposed delay constraints are oblivious to the medium access control (MAC) layer and hence cannot take necessary measures. Thus, there is a need to devise a control mechanism that can translate the delay requirements to parameters that the network protocols can understand. Moreover, the delay control needs to be considered in the routing layer, as well. However, due to the dynamics of the routes and network traffic patterns, it is difficult to predict or control the delay. Nonetheless, parameters like node buffer (queue) length can be monitored which can help in making decisions that would avoid congested areas and hence lower the delay. It is yet crucial that routing strategies address the network performance with energy budget constraints; energy efficiency is thus an important criterion in ad hoc routing protocol design. Further, there may be situations that both delay control and energy efficiency is desired, and the network should be able, by carefully designed schemes, to understand and operate in a proper way to meet the demands.

In this paper, we provide a generic interface framework that translates user supplied inputs into strategies that nodes in the network can adopt to satisfy the user requirement. We formulate the problem of user input driven QoS management and propose a design solution by using translation functions. We illustrate how the translation functions work, and offer case studies on MAC and routing layers in an IEEE 802.11 based ad hoc network. We take delay as an input, analyze the MAC functionality and show how to translate the delay into node transmission (channel access) probabilities. We validate the translation process by employing a similar slotted aloha model and discuss how carefully choosing the backoff timer helps to maximize the probability of achieving a target delay. Further, we derive the associated cost translation function with respect to the energy consumption for a given delay input. For the purpose of routing, we consider two inputs that the user can specify: delay and lifetime. These inputs translate to routing strategies that achieve the corresponding objectives. Moreover, we extend the model to incorporate both pure best effort and mixed strategies. To show the functionality of the proposed interface, we conduct ns-2 based simulations and demonstrate how the network QoS performance changes when different user inputs result in different strategies at both the MAC and routing layers.

QoS provisioning has been in the forefront of networking research. To address the QoS support and management issues, many models and middlewares have been proposed, for example in [20], a negotiation and adaptation mechanism is implemented. There is a rich set of the literature that addresses the QoS issues and their attainability in ad hoc networks (see [19, and the references therein]). However, there is always a pursuit to integrate multiple QoS metrics– all of which are correlated with each other. Typical metrics of interest are packet delay, energy consumption, throughput, and network lifetime. Each of the QoS metrics has different manifestations at various layers of the protocol stack.

To understand how QoS gets affected in MAC, modeling the MAC protocol is of utmost importance. Since the seminal work of Bianchi [2], studies on the behavior of IEEE 802.11 distributed coordination function (DCF) have provided insights on QoS provisioning in both IEEE 802.11 based WLAN and ad hoc networks. Extensive work has been done to model, analyze, and tune protocols to reach maximum system capacity [4], [5], [12], [18]. To better support QoS, IEEE 802.11e [14] has been proposed that enhances the performance by defining traffic classes (TCs). Some delay-sensitive applications, like voice over IP (VoIP), benefit from the new standard as these delay-sensitive applications could be assigned a higher priority class and thus experience less delay [24]. However, controlling the delay at the MAC layer of IEEE 802.11 is still an open problem, especially when the network operates in an ad hoc mode. In [6], the authors focus on the contention window and analyze delay characteristics in one-hop scenarios. A similar model is used in [11]; however, the new approach is extended to IEEE 802.11e MAC and the expected delay for different TCs are presented. An unique way to control the contention mechanism in IEEE 802.11e MAC is presented in [13]. Although the aim of the approach is to increase the throughput, the same method can be adopted for delay control as well. Also, MAC protocol design with energy consideration [15] is an effective way for QoS provisioning.

As far as QoS provisioning through routing is concerned, delay sensitive and energy efficient techniques have always been of interest given a limited bandwidth and battery budget. Xue and Ganz [25] focus on the end-to-end delay control and propose AQOR which uses traffic estimation. The bandwidth estimation method in [8] shows that packet delay and energy dissipation can decrease significantly by selecting a QoS-aware route. Bambos [1] introduced the minimum power routing protocols for the sake of energy conservation, while in [23], the authors proposed the idea of energy efficient routing with power-aware energy metrics. Similar metric based routing protocol was also proposed in [21]. Chang and Tassiulas [7] introduced a model in which energy efficient routing problem can be regarded as an optimization problem with the performance objective to maximize the lifetime of the batteries. Toh [22] proposed a routing protocol to maximize battery life by minimizing transmission power and evenly distributing power consumption among the nodes. Moreover, joint routing, link scheduling, and power control schemes [9] have also been considered to address the energy efficiency in ad hoc network routing.

The research in this paper stems from these advances in QoS provisioning at MAC and routing layers but differs significantly from existing work. Specifically, we are more interested in how user can be involved in the QoS provisioning process by using an QoS management interface. In designing the interface, existing models are employed to translate user inputs to MAC and routing schemes, however, modifications are made so that the strategies the network adopt will be adherent to the user inputs, instead of squeezing at the best level of QoS.

The rest of the paper is organized as follows. In Section 2, we formulate the user input driven QoS management problem and propose a possible interface to bring the user more closer to the network, helping the user to better manage QoS of the network. Demonstration of the interface and case studies with respect to MAC and routing layers are presented in Sections 3 Strategies and cost at MAC layer, 4 Strategies and cost at routing layer, respectively. We also show how user inputs can drive the networking strategies, with simulation model and results being discussed. Conclusions are drawn in Section 5.