Analyzing the effective throughput in multi-hop IEEE 802.11n networks

Abstract

In this paper we characterize the effective throughput for multi-hop paths in IEEE 802.11n based wireless mesh networks. We derive an analytical model capturing the effects of frame aggregation and block acknowledgements, features found in the new IEEE 802.11n standard. We describe the throughput at MAC layer as a function of physical data rate, error rate, aggregation level and path length. While being mathematically tractable, the proposed model is flexible enough to account for complex and realistic error characteristics of the wireless channel, such as long-term fluctuations and burstiness. We further show how to integrate the well-known Gilbert–Elliot channel model into our model and evaluate both models in our indoor wireless testbed.

Introduction

In recent years, wireless mesh networks became increasingly interesting in academia and industry because they can easily be built with low infrastructure costs despite a huge coverage. Therefore, they are particularly attractive for providing fast and cost-efficient coverage for hard-to-wire areas. Further areas of application include disaster scenarios, wireless machine-to-machine communication, and wireless video surveillance. The IEEE 802.11n standard [1] is the first IEEE 802.11 amendment to introduce a physical layer based on Multiple-Input and Multiple-Output (MIMO) transmission scheme, providing data rates up to 600 Mbit/s and increased tolerance to interference. These features make IEEE 802.11n a promising technology for building carrier grade wireless mesh networks. The high data rates provided by the IEEE 802.11n physical layer can only be harnessed at upper layers if medium access is efficient. Therefore, IEEE 802.11n introduces frame aggregation. Using frame aggregation, multiple subframes can be transmitted in sequence with the overhead for medium access arising only once. One option allows each subframe transmitted in an aggregated frame to be guarded by an own Cyclic Redundancy Check (CRC) checksum, i.e. the IEEE 802.11n Medium Access Control (MAC) at the receiver can extract individual subframes even if parts of the aggregated frame are erroneous due to lossy channel conditions. Upon reception of an aggregated frame, the receiver can send a BlockAck control frame to acknowledge all correctly received subframes.

For the design of data transport protocols, an understanding of the effective throughput of IEEE 802.11n is required. An appropriate analytical model has to consider the aggregation capabilities as well as partial retransmits occurring in IEEE 802.11n. Furthermore, the characteristics of the wireless channel as a scarce resource shared among all mesh nodes within their radio transmission range has to be considered.

In this paper we characterize the effective throughput for multi-hop paths in IEEE 802.11n based wireless mesh networks. We derive an analytical model based on a Markov chain capturing the effects of frame aggregation and block acknowledgements, features found in the new IEEE 802.11n standard. Our model calculates the expected number of retransmissions by estimating the amount of subframes to be retransmitted under a given channel model. With our Markov model, we are able to describe the throughput at MAC layer as a function of physical data rate, error rate, aggregation level and path length. Furthermore, we show how to integrate the well-known Gilbert–Elliot channel model into our model to take complex error characteristics into account. We develop a Markov chain counting the correctly received number of subframes in an aggregated frame of arbitrary size under the Gilbert–Elliot channel model. This way, while being mathematically tractable, the proposed model is flexible enough to account for realistic error characteristics of the wireless channel, such as long-term fluctuations and burstiness. Subsequently, we validate our model through simulations for multi-hop communication on a chain topology in IEEE 802.11n. We hereby show that our analytical model is able to closely estimate the effective throughput on these topologies.

Furthermore, we introduce our Indoor MIMO testbed and quantitatively show therein, that the Gilbert–Elliot channel model is able to describe error behavior in IEEE 802.11n communication. We employ the Baum–Welch algorithm to estimate the parameters of the underlying Gilbert–Elliot channel model and show that it is able to model error characteristics like mean subframe error rate, distribution of error burst length, and the distribution of the burst length of correctly received subframes in our testbed. Finally, we evaluate our proposed Markov model by estimating the throughput of different testbed links.

The remainder of this paper is organized as follows. Section 2 summarizes related work on analytical models for IEEE 802.11n and data transport optimization for wireless mesh networks, Section 3 gives a short overview regarding the frame aggregation and block acknowledgement features of IEEE 802.11n. In Section 4, we introduce our analytical model. In Section 5 we validate our model through simulations and evaluated it further in a real-world indoor testbed in Section 6. Finally, concluding remarks are given.