Studies on Cooperative Communication for Wireless LAN

(1)

Waseda University Doctoral Dissertation

Studies on Cooperative Communication for Wireless LAN

無線

LAN

における協調的通信に関する研究

February 2012

Graduate School of Global Information and Telecommunication Studies

Verotiana Hanitriniala Rabarijaona

(2)

To Papa, Maman,

Domdom, Pisokely

(3)

i

Summary

The last few decades have witnessed the exponential development of telecommunications worldwide. In particular, the demand for communication ubiquity drew the researchers’ focus on wireless communications. Being able to communicate anytime and anywhere was the main goal advertized by wireless communication technologies.

Similarly to people speaking to each other at the same time, wireless communication devices share the same medium to exchange information.

Collisions between two messages may thus occur. How to control the access of devices to the medium so as to avoid collisions is one of the fundamental problems of wireless communications. The most popular protocol developed to address this issue is the Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA) protocol, implemented in the well-known IEEE 802.11 standard.

CSMA/CA implements a Distributed Coordination Function (DCF) to administer the medium access.

Nowadays, the increasing presence on the market of electronic devices with wireless connection capabilities is driven by decreasing unit cost and enhancing portability and services. The consequences are a higher user density and mobility in wireless networks. This trend offers a new challenge compelling for improving scalability and adaptability for all layers of the wireless systems. As a result, new paradigms in system design have been sought for by the research community, notably leading to the notion of cooperation between network nodes.

Cooperation has been a topic of high interest for the last decade and has first been proposed to alleviate the wireless medium impairments. Indeed, the random nature of the wireless medium, also called wireless channel, causes random attenuations of the signal strength, called fading, which impairs the reliability of transmitting from one network node to another. This work focuses on the implementation of cooperation at the MAC layer of an 802.11 wireless network.

While cooperation was first explored as a new spatial diversity technique, we

(4)

address the issue of wasteful duplicate transmissions and investigate on-demand cooperative transmission. The objective is to provide cooperative relaying only when the original transmission fails. In contrast with the space diversity techniques where several copies of the same data are transmitted from different locations with users experiencing different levels of fading, we propose a protocol that leverages the diversity benefits for error recovery.

An overview of 802.11 MAC protocol is presented in Chapter 1 with the basic frame sequence of the CSMA/CA protocol and its behavior in case of error. The historical background of cooperation, its motivations and its basic principles are introduced. The fundamental cooperation algorithms are described and the aspects of cooperation at the MAC layer contributing to the diversity, the error recovery or the data rate improvement are investigated. Finally, we define the scope of this research and describe the constraints and the requirements to consider in a cooperative protocol.

Chapter 2 explores the potential contribution of cooperation at the MAC layer on a theoretical basis. A cooperative MAC protocol is proposed to reduce the overall latency of the network and to enable packet recovery after an error. The protocol introduces a new interframe space called Cooperation Interframe Space (CIFS) and a Request To Cooperate (RTC) control packet to perform cooperation.

Exploring the performances in an ideal channel and a saturated network sets the baseline regarding the expectations from a cooperative MAC protocol in more realistic scenarios. The theoretical analyses are carried out while assuming a fixed partnership, i.e. a station’s partner never changes; and the partners are assumed to be assigned randomly. Consequently a more adequate partnership selection can yield higher performances. Once the contribution of cooperation is confirmed theoretically with a random partnership, a low complexity SNR-based dynamic partnership is investigated and the cooperative MAC protocol is refined. It is shown that the use of cooperation can expand the coverage area of a terminal while maintaining a certain data rate.

Chapter 3 investigates a fuzzy logic based cooperative protocol, FuzzyCoop.

Researchers have studied the contribution of cooperation at the physical layer, especially focusing on outage probability and capacity regions; however the selection of an optimal partner to cooperate with remains a fundamental issue for efficient cooperation. Besides the inherent wireless channel impairments, the

(5)

iii number of users and the portability of the devices leaves crucial points to be addressed, namely the network density and the terminals’ mobility. Therefore, the implementation of a dynamic and adaptive partner selection becomes even more important. The proposed partner selection scheme involves parameters related to the channel instantaneous conditions, the transmissions history and the mobility of the terminals. The partner will be chosen based on a fuzzy logic algorithm. To ease the comprehension of the fuzzy logic-based partnership scheme, a brief introduction to fuzzy logic is first given. Since an optimal partner should share a good channel with both the source and the destination, a source-destination consensus based selection is introduced. Cooperation implies transmitting data for another terminal, in other words, a station uses its own resources while not directly benefiting from the transmission. To compensate for the resources utilization of a cooperative communication, a contention-free channel access is offered to the cooperating terminal as a reward. In addition to the low latency and low overhead packet retransmission, this reward comes as another incentive to cooperate. FuzzyCoop is compared to the low complexity protocol introduced in Chapter 2 and to the DCF function implemented in CSMA/CA. A fuzzy logic- based partner selection shows a higher successful cooperation ratio compared to a selection relying only on the SNR. FuzzyCoop contributes to the network’s performances when we vary the packet size, the packet generation rate, the maximum diameter of the area around the access point and the number of terminals.

Chapter 4 introduces the integration of cooperation in rate adaptive networks.

There is an increasing demand for high data rate due to the spreading of multimedia (voice, audio, video) enabled device. In order to maximize the data rate in accordance with the wireless channel conditions, rate adaptation comes as a potential solution. Nonetheless, the duration of transmissions with low data rate occupies the channel to the deficit of the terminals with high data rate.

Considering this issue and the improvements gained from cooperation in the previous chapters, the merging of a cooperative MAC protocol with rate adaptive techniques comes as the next logical step of this research. The proposed cooperative rate adaptation (CRA) protocol aims for an adequate partner selection, reactive transmission rate selection, low overhead protocol, and fairness. A rate selection based on the channel conditions is used and the cooperating partner selects the retransmission rate reactively according the received RTC. CRA eliminates the overhead added by FuzzyCoop for the consensus-based partner

(6)

selection in Chapter 3. It is evaluated in comparison with a non-cooperative and a cooperative rate adaptation schemes, namely Receiver-Based Auto Rate (RBAR) and Cooperative Receiver-Based Auto Rate (CRBAR). The results show that CRA outperforms both protocols and results in a tradeoff between the packet delivery ratio and the throughput. In particular, CRA helps to cope with the network’s density and the users’ mobility by maintaining good performances when the number of terminals or the stations’ velocity increase.

Chapter 5 sums up this thesis and concludes this work on cooperative MAC protocols for 802.11 networks.

(7)

v

List of Figures

Figure 1-1: Cooperation motivation ... 5

Figure 1-2: Basics of cooperation ... 6

Figure 2-1: Fixed Partnership-based cooperation ... 20

Figure 2-2: Cooperation for latency reduction ... 21

Figure 2-3: Cooperation for on-demand data retransmission ... 22

Figure 2-4: RTC frame format ... 22

Figure 2-5: Maximum Achievable Throughput for non-cooperative and cooperative protocol and related improvement ... 27

Figure 2-6: Saturation Throughput for non-cooperative and cooperative protocol considering the number of terminals, data rate and packet size ... 28

Figure 2-7: Comparison of Non-Cooperative and Cooperative Packet Recovery ... 31

Figure 2-8: SNR-based Partner Selection ... 32

Figure 2-9: Packet Delivery Ratio for Each Data Rate-Network Radius Configuration ... 35

Figure 2-10: Throughput for Each ... 35

Figure 3-1: Cooperative MAC protocol and NAV settings ... 40

Figure 3-2: Cooperation incentive ... 41

Figure 3-3: RTS frame format conveying the pre-selected partners' identity ... 42

Figure 3-4: RTC frame format conveying the selected partner's identity ... 43

Figure 3-5: Fuzzy System for Partner Selection ... 44

Figure 3-6: Symbol Error Rate vs. SNR with the 64-QAM Modulation ... 47

Figure 3-7: Consensus-based Partner Selection ... 54

Figure 3-8: Throughput vs. Packet Generation Rate ... 56

Figure 3-9: Average Packet Delivery Ratio vs. Packet Generation Rate ... 57

Figure 3-10: Average Packet Delivery Ratio vs. Number of Terminals ... 58

Figure 3-11: Successful Cooperation Ratio vs. Number of Terminals ... 59

Figure 3-12: Average Packet Delivery Ratio vs. Area Diameter ... 60

Figure 3-13: Average Packet Delivery Ratio vs Packet Size ... 61

Figure 4-1: Example of partner selection considering the channel quality ... 65

Figure 4-2: Transmission rate selection at the source ... 67

(11)

ix

Figure 4-3: Fuzzy System for a Low Overhead Partner Selection ... 68

Figure 4-4: Average Packet Delivery Ratio vs Number of Terminals ... 75

Figure 4-5: Throughput vs Number of Terminals ... 76

Figure 4-6: Average Transmission Delay vs. Area Diameter ... 77

Figure 4-7: Throughput vs. Area Diameter ... 78

Figure 4-8: Average Packet Delivery Ratio vs. Terminal Speed ... 80

Figure 4-9: Throughput vs. Terminal Speed ... 81

List of Tables

Table 2-1: Parameters of IEEE 802.11a ... 26

Table 2-2: Reception sensitivity, coverage area and network radius for -5dB link margin for different data rates using Cisco Aironet 802.11A/B/G Wireless CardBus Adapter [42] ... 34

Table 3-1: Inferences rules for FuzzyCoop ... 53

Table 4-1: Inference rules for a low overhead fuzzy partner selection ... 73

(12)

Chapter 1. Introduction

1.1 Wireless Communications and the CSMA/CA protocol

The demand for ubiquitous communication means increases with the proliferation of electronic devices such as laptops, smartphones, tablets…and led to a great expansion of wireless networks. Nowadays, users of all countries and especially in developed countries, expect to be able to communicate anywhere and at any time. The most popular standard used for wireless communication is the IEEE 802.11. This standard implements a Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA) protocol at the MAC layer [1].

1.1.1 Frame sequence

The CSMA/CA protocol implements the Distributed Coordination Function (DCF). A terminal with a pending packet has to senses the channel idle at least for a DIFS (DCF Interframe Space) period. After this period, any terminals wishing to transmit decrements a random backoff counter. The backoff counter is a number of time slots randomly drawn within a [0, CW] interval, where CW is the contention window size and is defined as follows:

� CWmin < 𝐶𝐶𝐶𝐶 < CWmax

CW = 2^k– 1, k = 1,2, … (1-1)

(13)

Introduction 2 The terminal with the smallest backoff counter can transmit its packet once the backoff time is elapsed. This process is called contention and is performed after every transmission before beginning another one.

There are two access methods: the basic access where terminals send the data packet right after contention; and the RTS/CTS handshake access. In the RTS/CTS access, the user that wins the contention session sends a Request to Send (RTS) to the access point (AP) which replies with a Clear to Send (CTS) packet. The RTS/CTS handshake is followed by the data packet transmission. The DCF implements a positive acknowledgement with an ACK packet from the destination after a successful transmission. A Short Interframe Space (SIFS) is inserted between every frame transmission.

Every frame contains a “Duration/ID” field in its header. The information found in this field is used by the neighboring terminals to update a Network Allocation Vector (NAV) that predicts the channel occupation time. This avoids collision during ongoing transmissions.

1.1.2 Erroneous Transmission and Binary Exponential Backoff (BEB) algorithm

The frame sequence described in the previous subsection is carried out continuously by every terminal in the networks continuously as long as no error occurs. At the first transmission attempt of a packet, the contention window is set to the minimum (CWmin). Every time an error occurs, the source terminal of the erroneous packet has to go through the contention process again in order to attempt retransmitting the packet. Since the contention is performed with all the other users wishing to transmit, there is no guarantee that this terminal will win the contention and be able to retransmit the packet immediately.

On the other hand, the contention window size of the terminal generating the error is increased according to the Binary Exponential Backoff (BEB) algorithm. After each error, its contention window is doubled; more precisely the parameter k in equation (1-1) is incremented until a maximum CWmax is reached. As a result, the backoff counter is drawn from a larger contention window for the next contention, and the larger the contention window, the smaller the probability to win contention is as derived by Bianchi in [2]. A terminal with continuous bad channel

(14)

conditions may generate successive errors that lead to an exponentially increased contention window size. Consequently, terminals with impaired channels may have less transmission opportunities. The overall latency of the network also increases since an error is reported by the absence of an ACK from the destination within an ACK-Timeout period. The terminals are allowed to restart contention only once this ACK-Timeout is elapsed. In error prone channels, this latency may become significant and lead to decreasing performances.

(15)

Introduction 4

1.2 Cooperation Background

1.2.1 Motivation

Technology development during the last decades led to a great progress in the field of telecommunications in general and more particularly in wireless communication. In contrast with wired communication, the medium in wireless communication suffers from external disturbances that challenge the data transmission reliability. Mobile networks and the demand for ubiquitous connectivity add to the challenge of providing reliable communication.

Researchers have put in many efforts to provide communication systems able to overcome these issues and to satisfy the users demand.

Relay communication has been one of the methods proposed so far to address these challenges [3] [4There are at least three terminals in the relay network, namely the source, the relay and the destination. While the communication occurs between the source and the destination, the relay terminal assists the source in its transmission by relaying its data to the destination.

Diversity has also been explored to mitigate the channel impairments. The mobile radio channel suffers from fading, meaning that mobile users may experience variations in signal attenuation during a transmission. The severity of the attenuation is increased by the mobility of the terminals. The basic concept of diversity is to effectively transmit (semi)independently fading copies of the signal for directly counteracting the effects of fading. Some well-known forms of diversity are spatial diversity, temporal diversity, and frequency diversity [5].

Space diversity is of particular interest and has been investigated in [6][7][8].

There is often no line-of-sight between a source and a destination in wireless communication (indoor or outdoor). A signal is reflected on many different paths and suffers from fading before reaching the destination. Spatial diversity relies on the principle that signals transmitted from geographically separated transmitters, and/or to geographically separated receivers, experience fading that is independent.

(16)

Figure 1-1: Cooperation motivation

In contrast with the relay communication where the signal is retransmitted from a different transmitter, spatial diversity implements an antenna array on a single transmitter.

Cooperation is first studied by Sendonaris [9][10]. The work on cooperation is inspired on the benefits of relay networks and spatial diversity. In the relay channel, the sole purpose of a relay terminal is to assist a source in transmitting its data to a destination. In contrast, a cooperating terminal relays another terminal’s data but also has its own data to send.

Space diversity is another approach where co-located antenna array transmits redundant signals over independent channels. However, mobile terminals ranging from computers to mp3 players… are increasingly becoming smaller and it may not be always feasible in practice for each user to have multiple antennas.

(17)

Introduction 6

Figure 1-2: Basics of cooperation

Consequently, a new class of diversity technique called cooperative diversity has been proposed. Distributed users interact with each other to jointly transmit information exploiting diversity offered by multiple users. In order to overcome the terminals’ size limitation obstructing the multi-antenna implementation, a virtual antenna arrays is created between the users of a network.

Cooperation is motivated by the inherent omnidirectional feature of the wireless channel. Therefore when a source wants to communicate with a destination, there is no extra cost in sending the signal to the other terminals in the network (provided they lie within the transmission range of the source). Moreover, a signal transmitted from two different terminals experience different levels of fading.

These phenomena are depicted in Figure 1-1.

1.2.2 Basic Principles

The fundamental idea behind cooperation is that each user has a “partner” also known as “relay” or “helper”. Each of the two partners is responsible for transmitting not only their own information, but also the information of their partner, which they can overhear. Cooperation attempts to create spatial diversity through the use of the partner’s antenna.

There are two phases in a cooperative transmission as illustrated in Figure 1-2.

During the first phase, the source transmits its data which are received by the destination and overheard by the partner. The partner relays the received data to

(18)

the destination. Diversity is achieved with the two copies of the same data sent from different locations [9][10].

Several cooperative diversity algorithms have been investigated and multiple partners have been considered. The cooperation algorithms can be categorized into two types: repetition-based and space-time coded cooperation [11][12]. The repetition-based cooperation consists of the source terminal broadcasting its transmission to its receiver and to potential partners and the relays repeating the sender’s message individually on orthogonal channels [13][14]. Amplify-and- Forward (AF) and Decode-and-Forward (DF) are two well-known repetition- based techniques. AF allows the partners to amplify the signal received from the source and repeat its transmission. Partners fully decode the overheard signal before relaying it when DF is used. The corresponding benefits come at a price of decreasing bandwidth efficiency (increasing time delay) because each relay requires its own channel (time) for repetition. Space-time coded cooperation operates in a similar fashion except that all the relays transmit simultaneously on the same channel using a suitable coding scheme such as orthogonal distributed space-time code (DSTC) [15][16].

(19)

Introduction 8

1.3 Cooperation at the MAC Layer

As described in Section 1.2, cooperation has been inspired by relay communication. Originally, the research on cooperation focused on the physical layer but cooperation at the MAC layer has recently attracted attention.

1.3.1 Cooperative Diversity

Cooperation was first investigated as a new form of diversity also called cooperative diversity. After a major focus on the physical layer, cooperative diversity at the MAC layer recently drew researcher’s attention.

The protocol in [23] proposed a three-way handshaking to determine when to cooperate and whom to cooperate with. This handshaking involves a RTS, a CTS and RRTS (Relay Ready To Send) packet exchange between a source, a destination and potential partners. The source terminal sets a target rate conveyed in the RTS. Cooperative transmission is performed under two conditions: i) The direct link (source-destination) cannot support the target rate. ii) The cooperative link (source-partner-destination) can offer a greater rate than the direct link. The RTS and the CTS frames are modified to provide information enabling the neighboring terminals to evaluate these two conditions. A terminal that satisfies i) and ii) transmits a RRTS. In case there are more than one relay candidate, a contention mechanism using busy tones is carried out and prioritize the neighbors with capacities to reduce the collision probability two or more RRTSs. The protocol achieves diversity gain for different target outage probabilities. However it also introduces signaling overhead with the busy tones and the RTS, CTS modification. The latency is increased through the relay candidates’ contention.

Despite a prioritization scheme, there is a risk of collision, especially in high density networks.

In [24], a terminal chooses its relay based on the SINR of the packets received from its neighbors. The terminal informs the network of the identity of the chosen relay in the unused address field of a DATA packet or in a Hello packet if it has not sent any DATA for an extended period of time. Cooperative diversity is activated if a transmission on the primary link fails, i.e. if a source does not

(20)

receive a CTS within a CTS Timeout period, a C-RTS (cooperative RTS) is transmitted cooperatively. In other words, the source and its relay transmit the same copy of the C-RTS. In the same way, a C-CTS (cooperative CTS), a C- DATA (cooperative DATA) and a C-ACK (cooperative ACK) are transmitted cooperatively by the destination and the source and their respective partners. The protocol increases the transmission reliability through cooperative diversity by transmitting duplicate copies of all the packets from the source, the destination and their respective partners. However, the timeliness of the partner selection is not addressed, i.e. the relay is selected proactively and channel changes between the relay selection and the actual cooperation are not considered. Control packets sizes are usually smaller and transmitted at a lower rate than data packets and they are less error prone than the data packets. Therefore duplicate transmission of control packets may become a waste of resources.

Neighboring terminals of a source S and a destination D compete for cooperation based on the CSI (Channel State Information) between them and S and D respectively in [25]. The neighbors estimate the CSI values by monitoring the RTS from the source and the CTS from the destination. After the RTS-CTS handshake, the relay candidates decrease a timer inversely proportional to the CSI and transmit a AFR (Apply For Relay), i.e. the neighbor with the best CSI is prioritized in the contention process. D replies with a SFR (Select For Relay) to inform the network of the chosen relay. Cooperation is performed whenever an ACK is not transmitted, i.e. when the packet has not been received successfully by the destination. The relay selection is skipped to reduce the power consumption whenever the PER between S and D is below a threshold. On the other hand, the neighbors retreat from the relay selection process when the PER between them and S or D is above a certain threshold. Despite reduced outage probability and energy savings, the protocol introduces new control packets that do not exist in the CSMA/CA protocol. The cooperation decision is also proactive and do not depend on whether the data packet is received correctly or not.

The relay is distributively selected according to the instantaneous estimation of the wireless channel condition upon reception of the RTS from the source and the CTS from destination in [26]. The neighbors decrement a timer inversely proportional to the end-to-end channel condition. The selected relay is the one whose timer expires first and it informs the network with a flag packet. The data is

(21)

Introduction 10 transmitted in two steps; the source sends the data in the first step and the packet is relayed by the partner in the second step. The proposed method avoids the complexity incurred by space-time coded cooperation without reducing the performances. The protocol purely aims for diversity and do not consider wasteful duplicate transmissions in case the first transmission succeeds.

In [27], diversity is used for the data frames but also for the control frames.

Instead of the regular RTS, a cRTS (cooperative RTS) is used to inform about the chosen relay. The subsequent packets (CTS, Data and ACK) are transmitted twice by the source or the destination and a common relay. The protocol is evaluated through experiments implementing three terminals (source, relay and destination) to measure the performance of the cooperative WLAN in indoor and vehicular scenarios. The results show improvements in the throughput and the packet error rate. However the experiments stage three terminals only and the relay selection is not addressed. Besides, the duplicate transmissions of every packet may be wasteful when the channel does not suffer from severe fading.

1.3.2 Cooperative Error Recovery

Some protocols avoid the wasteful multiple transmissions by enabling cooperation only when a transmission error occurs, i.e. for error recovery.

A distributive relay selection based on the work in [23] is proposed in [28].

Offner and Adam leverage cooperation for error recovery, calling a partner only if a packet is not received successfully. The destination estimates the Packet Error Rate (PER) upon reception of a RTS from a source. If it falls below a threshold, the destination informs the network that a relay might be needed by transmitting a CCTS (Conditional CTS) packet. Relay candidates signalize their presence by using busy tones. The source consults the busy tones to estimate the number of candidates and report that number in a NRC (Number of Relay Candidates) packet.

The competing relays use this number to derive the size of their contention window. The terminal that wins contention sends a RRTS (Relay Ready To Send).

After the relay contention, the source finally transmits its packet which is overheard by the relay. If the selected relay does not detect an ACK after the data packet transmission, it retransmits the packet. The results show an increase of the

(22)

throughput and a decrease of the dropped packets. However, the wait for the absence of the ACK generates latency in the network; besides the relay selection is proactive and becomes a resources waste when the data transmission is successful. Moreover, the signaling for cooperation (busy tones, NRC) adds overhead to the protocol.

A cooperative CSMA/CA (C-CSMA/CA) is introduced in [29]. The frame sequence follows the standard CSMA/CA with a RTS/CTS handshake and the DATA transmission. The neighboring terminals overhear the packet. Cooperation comes in only when an error occurs, i.e. if no ACK is received. The idle terminals that overheard the packet, as well as the source, take part in the next contention session for its retransmission. In order to give transmission opportunity to the terminals with good channel and to carry the cooperative retransmission as fast as possible, a prioritization scheme based on the Channel to Noise Ratio (CNR) and the transmission objective (first transmission, retry, relay) is implemented. The C- CSMA/CA protocol provides error recovery and improves the throughput and the delay of the network. However despite the prioritization scheme, contention among the relays candidates may cause collision with greater risks for a crowded network, making the system unable to guarantee immediate packet retransmission.

Cooperation becomes difficult to perform in heavy traffic condition where the terminals are not idle.

As in [29], if no ACK is received after a packet transmission, the terminals that decoded the frame correctly proceed to cooperative retransmission in [30]. More than one terminal can cooperate to achieve diversity. To avoid collision, the partners choose a backoff time within [0, CW^p] where CW^pis the number of associated stations announced by AP (Access Point) in a beacon. A combination of cooperation with the Forward Error Correction (FEC) technique where the data frame is split into smaller blocks is also proposed. In spite of higher performances compared to the non-cooperative protocol shown by the numerical analyses, the protocol does not address the relay selection which is one the main issue of cooperation.

(23)

Introduction 12

1.4 Cooperative MAC protocols and data rate improvement

1.4.1 Introduction to rate adaptation

Adapting to highly varying wireless channel condition is very challenging for the terminals. One of the solutions explored in the literature is known as rate adaptation and works with the 802.11 standard which offers several physicals layers supporting multi-rate capabilities. Rate adaptation techniques select the transmission rate according to the varying link condition.

Auto Rate Fallback (ARF) was the first rate adaptation algorithm proposed [31].

The rate is increased or reduced after a certain number of transmission successes or failures respectively. It uses the transmission history to set the transmission rate and as a result is not an algorithm that adapts to the channel fluctuations.

Receiver Based Auto Rate (RBAR) comes as a more adaptive scheme to the channel alteration [32]. In RBAR, the source begins the communication with an RTS conveying its predicted transmission rate and the packet size instead of the duration field. Upon reception of the RTS, the receiver estimates the SNR and selects the transmission rate accordingly. Then, the receiver transmits the selected transmission rate in the CTS. The source terminal chooses the highest rate between its predicted rate from the RTS and the one calculated by the destination found in the CTS. The source includes the rate selection in a Reservation SubHeader (RSH) of the data packet. Finally, the receiver ends the communication with an ACK. A threshold-based technique is used for the rate selection algorithm. The chosen rate is the highest transmission rate that allows a bit error rate (BER) lower than 10⁻⁰⁵.

Let M1, …, MN be the modulation schemes corresponding to the available transmission rates; θ1, …, θN be the SNR threshold at which BER(Mi) = 10⁻⁰⁵; and i = 1, … , N. The selected modulation scheme is presented in equation (1-2).

(24)

�M₁ if SNR < θ₁ M_i if θ_i< 𝑆𝑆𝑆𝑆𝑆𝑆 < θ_i+1

MN otherwise (1-2)

1.4.2 Cooperation applied to rate adaptation

The promising results against channel variation offered by rate adaptation and diversity and the increased reliability provided by cooperation inspired the researchers to investigate the implementation of cooperation into rate adaptive networks.

The CoopMac protocol is proposed in [33]. Every stations holds a table of its neighbors ID, the last time when a packet has been received from them, the data rate between them and the AP (Access Point) and the data rate achievable between themselves and their neighbors. The chosen helper is the neighbor with the best source-helper-destination data rate. Two versions of the protocol are proposed:

CoopMac I and CoopMac II. In CoopMac I, the RTS conveys the helper’s ID and the expected data rates. A HTS (Helper ready To Send) is used by the helper to inform the source of if it can maintain the expected data rates. The packet is transmitted cooperatively if it can be delivered faster in two hops (source-helper- destination) rather than through the direct link. To avoid the RTS modification and the use of HTS, CoopMac II proposes to convey the helper’s ID in the fourth address field of the data frame. It reduces the overhead incurred by CoopMac I but is more vulnerable to the channel changes since the chosen helper cannot confirm its availability to cooperate.

To increase the data rate transmission, the terminals are divided in groups according to their achievable data rate as described in [34]. When a terminal is in the highest rate group, cooperation is not used. Otherwise AP replies to the RTS with a cCTS (cooperative CTS). The relay candidates are the stations that are within the transmission range of the source and AP and are in a group with a rate higher than the source. They send a riCTS to the source after a random backoff time. When a relay candidate overhears a riCTS, it sends its own riCTS if the data rate it can provide is higher than the data rate specified in the riCTS. Otherwise it remains silent. The source indicates the suitable relay’s identity in the unused address field of the data frame. The relay acknowledges the packet and sends it

(25)

Introduction 14 again together with its own data at his group data rate. The protocol increases the overall transmission rate of the network; nonetheless it adds heavy signaling overhead with the riCTS and time overhead through the contention between the relays.

The work of [29] is extended in [35] where cooperation is incorporated into rate adaptive networks. It proposes to reduce the low rate transmission by using the highest rate for every first retransmission. The packet can be retransmitted several times at high data rates. However after Mtimes retransmission failures, the source reverts to the rate adaptation mode for next retry. A prioritization scheme sets the Arbitration Interframe Space (AIFS) of each terminal according to their CNR and the purpose of their transmission (first try, retry, relay). Despite the prioritization of the terminals with good CNR, the terminal with the highest CNR does not necessarily support the highest transmission rate. Contention must be performed at every transmission retry or relay adding to the overhead and the latency is increased with the wait for the ACK.

A Cooperative Receive Based Auto Rate (CRBAR) scheme is proposed in [36].

Communication begins with a RTS containing the packet length from the source.

The destination replies with a CTS. The neighbors estimate the SNR of the RTS and the CTS. They use this information to determine the source-neighbor transmission rate and the neighbor-destination transmission rate. The transmission rates are determined in the same way as in [32] described in equation (1-2). The neighbors use the packet length and the selected transmission rates to calculate the time needed for a transmission without relay (direct transmission time) and the time needed for a transmission with a relay (cooperative transmission time). If the cooperative transmission time is shorter than the direct transmission time, the neighbor will contend to send a Ready to Relay (RTR). However, short transmission duration does not ensure a successful communication and contention among the relay candidates increases the delay and may induce collisions.

(26)

1.5 Scope

Cooperation is first investigated as a novel diversity technique and is studied extensively at the physical layer. Cooperative diversity algorithms have been developed and capacity theorems have been explored [9]-[22]. This thesis will focus on the study of cooperation at the packet level in the MAC layer. The work will be based on the 802.11 standard which is currently the most widespread wireless communication standard.

The original purpose of cooperation was to achieve diversity. The possibility of wasteful multiple transmission is addressed in this thesis. Instead of aiming at providing diversity as numerous works throughout the literature have already done, this thesis leverages the wireless medium’s omnidirectional nature and the diversity capability of the virtual array embodied by the terminals to strengthen the transmission reliability. This is achieved by offering error recovery to improve the overall network performances and more particularly, to reduce the numbers of dropped packets.

An increasing number of electronic devices can now be connected to a wireless networks as technological advances have made them smaller and more portable.

As a result, mobility becomes a crucial issue in the communication performance.

Moreover, wireless communication is becoming available in crowded places such as shopping malls, coffee shops, train stations and the like, increasing the number of users and the demand considerably, and leading to the exponentially increasing amount of data being exchanged every day. Therefore maintaining good performances in high traffic and dense networks is a challenge that still needs to be addressed, and the selection of the partner for cooperation is crucial.

The contribution of cooperation at the physical layer has already been stated and before delving into the partner selection issue, a theoretical study will be carried out on the contribution expected from cooperation implementation into a wireless network.

The wireless channel is prone to variations due to the stations mobility or the movement around the stations (vehicles, door closing, moving obstacles).

Furthermore, text is no longer the only type of the data being transmitted; new types of data such as images, audio, voice and video require higher data rate

(27)

Introduction 16 transmission. Rate adaptation has been considered to overcome the signal attenuations while maintaining a certain transmission rate. The incorporation of cooperation into rate adaptive systems is also studied in this work.

(28)

1.6 Constraints and Requirements

Although cooperation was initially proposed as a diversity technique, cooperative retransmission should not be systematic. It may be possible that the first transmission succeeds, depending on the communication conditions, in which case any subsequent retransmission becomes a waste of the medium. Therefore, a reactive cooperative protocol should be able to offer on-demand relay only when an error occurs.

The wireless medium being a scarce resource, a protocol must provide an efficient usage of the channel minimizing the overhead while maximizing pure data transmission. The CSMA/CA protocol comes with a set of overhead (control packets, header, interframe space) that is essential for a smooth communication.

Although the number of control packets can be reduced by eliminating the RTS/CTS handshake, it is advisable to use it to avoid collision between hidden nodes especially when transmitting large sized packets or when the network is crowded. Henceforth, it is desirable to create a cooperative protocol that can exploit the available information in the existing overhead of CSMA/CA to prevent adding signaling overhead such as busy tones or control packets from the relay candidates.

Time overhead should also be reduced as much as possible. When an error occurs, time overhead or latency can have two origins: the wait for the ACK timeout and the contention period between the partner candidates. Latency is all the more significant when no relay is available and the contention period becomes merely a waste of time. Consequently, cooperation should minimize as much as possible the latency. This can be achieved by refraining from performing contention among the neighboring terminals or even by eliminating the whole process. The network should be notified of an error immediately after its happening so that the stations do not have to await the ACK timeout.

A standard frame format also comes along with CSMA/CA. When adding new frames to a cooperative MAC protocol, those frames should abide by the standard format in order to facilitate its incorporation into the legacy protocol.

(29)

Introduction 18 As explained in section 1.1.2, when a terminal transmits a packet that is not acknowledged, it will join the following contention session with a doubled contention window size. As a result its probability to win the contention diminishes accordingly. When a station suffers from dire channel conditions for an extended period of time, it may be prone to successive errors, yielding an exponentially increasing contention window size and reducing its opportunity to transmit. This phenomenon is at the origin of a fairness issue where terminals sharing a good link with the destination benefit from more transmission opportunities at the expense of those with impaired channels. Besides, contending for the channel after an error does not guarantee an immediate retransmission since other terminals with smaller contention window are also competing to transmit. Ergo, a cooperative system addressing the fairness issue and ensuring a prompt packet relay should be investigated.

On the other hand, rate adaptation has been introduced to allow the terminals to adapt its transmission rate to the channel’s fluctuations. However terminals with bad links will communicate at low data rates and monopolize the medium for a long period of time. This phenomenon becomes in turn unfair to the stations with good channel that do not requires a long transmission time with higher data rates.

Hence, it is required from the cooperative protocol to allow faster transmissions at higher rates while maintaining reliability.

Finally, due to the spreading of wireless technologies and to their affordability, the numbers of users is evermore increasing. Moreover, thanks to the portability of recently developed electronic devices, the users can move easily while transporting them. In consequence, network density and stations’ mobility are two challenges to be tackled by cooperation.

(30)

1.7 Thesis Outline

This thesis is divided into six chapters:

Chapter 1. gives the background of cooperation, the motivation and the basic concepts. It offers an overview of the CSMA/CA protocol used in the well-known IEEE 802.11 standard. The cooperation methods at the MAC layer reported across the literature, an introduction to rate adaptation are presented. The merge of cooperation protocols with rate adaptation and their limitations are discussed. This chapter also states the constraints and requirements of a cooperative protocol and defines the scope of this work.

Chapter 2. investigates the theoretical contributions expected from the implementation of a cooperative MAC protocol in 802.11 networks. It exposes the limitations of a fixed partnership and introduces a dynamic partnership. The extension of the stations’ coverage area through cooperation is explored.

Chapter 3. describes a novel fuzzy logic algorithm for partner selection relying on the error history, the instantaneous channel conditions and the coherence time. It proposes a source/destination consensus-based dynamic partnership. An incentive to cooperate and a compensation scheme for the utilized resources for a cooperative retransmission is examined as well.

Chapter 4. focuses on the implementation of cooperation in rate adaptive networks and presents its contribution towards the fairness of the network. A reactive rate selection by the partners is investigated. We analyze the performances while considering the mobility and the density of the network.

Chapter 5. summarizes this thesis and concludes our work.

The list of publications ensuing from this work and the references are added at the end of this thesis.

(31)

Investigation on the Cooperation’s Potential in Wireless Networks 20

Chapter 2. Investigation on the

Cooperation’s Potential in Wireless Networks

2.1 Fixed-partnership cooperation

This chapter introduces a partnership-based Cooperative MAC protocol. It is assumed that each user has a partner and the partnership is fixed, i.e. the partners remain the same at anytime. Let us consider the partnership of the terminals (T1, T2) as shown in Figure 2-1. The cooperation between T1 and T2 will aim to reduce the latency and to provide immediate packet retransmission whenever an error occurs.

Figure 2-1: Fixed Partnership-based cooperation

(32)

Figure 2-2: Cooperation for latency reduction

2.1.1 Latency Reduction

The first objective of the proposed cooperative MAC protocol is to reduce the latency of the network. To do so, we introduce a new type of Interframe Space called Cooperation Interframe Space (CIFS).

Let us consider the T1-T2 partnership depicted in Figure 2-1. Assuming that T1 won a contention session, it sends a RTS to the Access Point (AP). If at the same moment its partner T2, has a pending packet, it is allowed to send its own RTS after the CIFS. The AP gives priority to T1 since its RTS was received first. After the ACK frame to T1’s data, the AP sends a CTS to T2 in response to its RTS received after the CIFS. This frame sequence is depicted in Figure 2-2. Doing so prevents T2 from channel contention if its partner already won the channel access, hence reduces latency. Since the channel is already reserved for the duration of both terminals’ communication, the risk of collision is alleviated for T2’s transmission. On the other hand, if T2 does not have any pending packet, after the CIFS, the AP waits for another SIFS before sending a CTS to T1. A CIFS period is defined by (2-1).

CIFS = SIFS + Time Slot (2-1)

(33)

2.1.2 On-demand Cooperative Packet Retransmission

The next and main goal of the protocol is to provide packet recovery after an error.

Communication begins with a RTS/CTS handshake between T1 and AP. Then T1 transmits its data packet. The omnidirectional feature of the wireless channel allows the partner T2 to overhear this packet. If the packet is not received successfully by AP, the latter sends a Request to Cooperate (RTC) to T2 to trigger packet recovery. T2 retransmits T1’s overheard packet and the AP replies with an ACK. This frame exchange sequence is depicted in Figure 2-3.

The RTC is a control frame that asks T2 to relay T1’s erroneous packet. The RTC frame format is defined as a CTS frame with an additional “sequence control”

field to identify the lost packet. The field “source address” identifies the terminal which initiated the transmission. The RTC frame format is shown in Figure 2-4.

As described in Section 1.1, in the standard CSMA/CA, the terminal generating the error must wait for an ACK Timeout and proceed to channel recontention to retransmit an erroneous packet. In this protocol, the error recovery is initiated on- demand as soon as the error occurs.

Figure 2-3: Cooperation for on-demand data retransmission

Figure 2-4: RTC frame format

(34)

2.1.3 Numerical Analysis

Maximum Achievable Throughput

The first metric to evaluate the performances of the proposed cooperation MAC protocol is the Maximum Achievable Throughput (MAT). To do so, a saturated network in an ideal channel with no collision and no error is assumed. The terminals are assumed to have the same data rate and the same packet size. The MAT is given in (2-2)[37][38]:

MAT =_D^L (2-2)

with

L: packet size

D: transmission delay of L

The transmission delay comprises all the idle time, namely the inter-frame spaces such as DIFS and SIFS (TIFS) and the backoff, the control frame (RTS, CTS, ACK) and the data frame transmission times (TCTRL and TDATA respectively). Thus, the transmission delay is as in (2-3).

D = TIFS + TCTRL + BO + TDATA (2-3)

BO is the backoff counter and is given in (2-4).

BO = ^CW^min₂^×T^slot (2-4)

with

CWmin : minimum contention window Tslot : time slot

(35)

Investigation on the Cooperation’s Potential in Wireless Networks 24 A data frame transmission time is formulated as follows:

T_DATA = T_PHY + T_SYM ^16+6+8×(L_N ^MAC^+L^DATA ⁾

DBPS (2-5)

with

TPHY : transmission time of the PHY layer overhead LMAC : MAC layer overhead

LDATA : Packet payload size

NDBPS : number of data bit per symbol depending TSYM : transmission time of a symbol at the PHY layer

The MAT represents the upper bound of the achievable throughput. It will give a snapshot of the potential contribution of cooperation in CSMA/CA-based wireless networks.

Saturation Throughput

The next step of the numerical analysis is to derive the saturation throughput considering the probability of collision and the channel error in a saturated network. First, we define the probability of collision. In the 802.11 protocol, collision may occur with a probability PC when two stations pick up the same backoff window size depending on the contention window size and the number of terminals. The probability of collision is formulated by (2-6) [39].

P_C = 1 − �1 −_1−P^2(1−2P^C⁾

C−P_C(2P_C)^m 1

CW_min�^N−1 (2-6)

where N is the number of terminals and we have m such as CWmax=2^mCWmin. A collision occurs at the transmission of the first frame in the sequence. Since the NAV is automatically updated after a successfully transmitted RTS, any

(36)

subsequent error is a channel error as derived in the following. A channel error occurs when there is no collision and the packet is not received correctly. This error has a probability PE where PER is the Packet Error Rate:

PE = (1 − PC)PER (2-7)

On the other hand, a packet is successfully received if there is no collision and no channel error. The probability of success is:

P_S = (1 − P_C)(1 − P_E) (2-8)

Using PS, PC and PE, we derive the average packet length E[L] and the average transmission delay E[D] given by equations (2-9) and (2-10) respectively.

E[L] = P_S × L (2-9)

E[D] = PC× TC + PS × TS × PE × TE (2-10) where TC, TS, TE are the time elapsed in case of collision, successful transmission, and error, respectively.

The saturation throughput becomes:

S =

_E[D]^E[L] ^(2-11)

(37)

2.1.4 Results

The numerical analyses are carried out using the parameters of CSMA/CA in the IEEE 802.11a standard found in Table 2-1. It is assumed that the network is saturated and the terminals transmit at a data rate of 12 Mbps or 54 Mbps.

Maximum achievable throughput

Figure 2-5 compares the MAT of the non-cooperative CSMA/CA and the proposed cooperative MAC protocol with respect to the data payload size as derived earlier. The right axis shows the percentage of improvement in the MAT that we can obtain with cooperation. Using cooperation allows reducing the time overhead especially in a saturated network. Thus, it helps to improve the throughput. We can see that cooperation improves the MAT with bigger packet payload size. This is because the channel utilization for real data increases. As shown in equation (2-2), the MAT is the ratio of the data payload and the delay. So the bigger the packet size is, the higher the MAT becomes. We can see that cooperation performs better than the non-cooperative CSMA/CA in low or high data rates. The improvement increases with the data transmission rate.

Table 2-1: Parameters of IEEE 802.11a

Parameters Value [1]

Time slot 9μs

SIFS 16μs

DIFS 34μs

CIFS 25μs

TPHY 24μs

TSYM 4μs

ACK_TO 50μs

CTS_TO 50μs

CWmin 15

CWmax 1023

LMAC 28 bytes

NDBPS (resp. OFDM-12, OFDM-54) 48, 216 DBPS

TRTS (resp. OFDM-12, OFDM-54) 36, 24μs

TCTS=TACK (resp. OFDM-12, OFDM-54) 32, 24μs

(38)

From equations (2-3) and (2-5), we can see that the transmission delay depends on the data transmission rate. A higher transmission rate leads to a lower transmission delay as implied by equation (2-2). As a result, the MAT increases as well.

Saturation Throughput

In Figure 2-6, we analyze the saturation throughput with respect to the number of terminals with different data transmission rate (12 and 54 Mbps) and with different packet size (500 and 1000 bytes). As recommended in [1], a packet error rate of PER=8% is assumed. We can see that the throughput decreases as the number of terminals increases. This is because the probability of collision rises with the number of stations, hence increasing the transmission delay E[D].

However, the cooperative protocol still gives a better saturation throughput in all situations. This improvement comes from the immediate packet recovery that eliminates the time latency due to the contention for packet retransmission.

Cooperative retransmission also has the advantage of reducing the control packets overhead (RTS, CTS). As in Figure 2-5, a higher data transmission rate and a larger packet size lead to greater improvements.

Figure 2-5: Maximum Achievable Throughput for non-cooperative and cooperative protocol and related improvement

(39)

Figure 2-6: Saturation Throughput for non-cooperative and cooperative protocol considering the number of terminals, data rate and packet size

2.1.5 Conclusion and Limitations

This section introduced a fixed partnership-based cooperative MAC protocol for wireless networks and evaluates the protocol in the 802.11a environment. This protocol aims to reduce the overall latency of a network. It also aims to leverage diversity from a partner to relay a packet whenever an error occurs. A new interframe space called Cooperative IFS (CIFS) is presented. It allows two partners to access the channel if one of them wins the contention. This keeps a partner from contending for the channel in another session if it has a pending packet to send at the same time, and therefore reduces the latency of the overall network. A control frame called Request to Cooperate (RTC) is also introduced. It allows the AP to ask the partner for help in case the original packet transmission fails. Through numerical analysis, we see that the proposed cooperative MAC protocol helps to increase the achievable throughput as well as the saturation throughput at any data rate, for any packet size.

(40)

This section sets the theoretical contribution expected from implementing cooperation in wireless networks. From the results, we can conclude that cooperation is a promising way to obtain higher performances in wireless networks. This theoretical investigation is the basis for further studies, analyses and simulations considering more realistic scenarios. In particular, fixed partnership is not suitable in a mobile network where the channel conditions change through time. Moreover, partnership assignment remains an issue to be addressed. A poor partner assignment may results in successive transmission errors from a partner with a channel worse than the source terminal. The use of CIFS may increase the latency, especially when the partner does not have a pending packet, in which case the destination waits an extra SIFS before replying with the CTS. In case the partner and the destination do not share a good channel, the second RTS from the partner may not be received successfully by the destination. These issues will be addressed in the following sections.

(41)

2.2 SNR-based dynamic partnership

A random partnership does not guarantee a successful cooperation in error prone channels. Moreover, a fix partnership is not adequate in a mobile network where the optimal partner can change with time. This section proposes a dynamic partner selection scheme for cooperative error recovery.

2.2.1 Error Recovery Process

As in the standard CSMA/CA, every station has to perform contention in order to win the access to the channel. The terminal that wins contention exchanges a RTS/CTS handshake with the destination and then transmits its data packet. If the data is not received successfully, the destination notifies the network with the Request To Cooperate (RTC) control frame. The RTC is used to inform the network that a packet has been corrupted at the reception. It allows avoiding the latency incurred by the waiting of an ACK timeout after an error. On the other hand, some protocols suggest performing cooperation in absence of an ACK [28][29][30][35]. Besides the resulting latency, another drawback of this method is that a terminal hidden from the destination that does not hear the ACK might attempt to retransmit a packet that has already been received successfully. The RTC helps to overcome this hidden terminal problem. The RTC’s second and main purpose is to trigger cooperation and to demand a cooperative packet retransmission from the partner. The packet is acknowledged if the cooperation is successful. This process is depicted and compared to the standard CSMA/CA in Figure 2-7.

Note that the use of CIFS is dropped in the refined cooperative MAC protocol.

The CIFS was introduced to allow a pair of terminals to win the channel access if one of them won the contention. Since the partnership is dynamic i.e. not fixed anymore, a terminal contends for the channel for itself only and not for its partner.

Moreover, Section 2.1 analyses the performances in a saturated network whereas in a realistic scenarios, the terminals do not necessarily have data to send. In that case, the CIFS merely adds to the latency of the network.

(42)

Figure 2-7: Comparison of Non-Cooperative and Cooperative Packet Recovery

2.2.2 SNR-based Partner Selection

Although the protocol offers on-demand cooperative packet relay, a crucial question remains: who to cooperate with? To answer this question, each terminal holds a Potential Partner Table (PPT) containing the IDs of the neighboring terminals potentially able to cooperate. Every time a terminal overhears a packet successfully, it estimates the SNR. Then, it inserts the packet sender’s ID into the PPT in SNR order, i.e. the greater the SNR, the higher the neighbor is in the list. It is important to note that only the terminals whose packets are successfully received are inserted in the table. This is because if communication fails between two stations, cooperation between them is also likely to fail. When a terminal, transmits a RTS (Request To Send) frame, it uses the fourth unused address field to inform the Access Point (AP) of the selected partner in case of error. This selected partner corresponds to the first terminal listed in the PPT.

(43)

Figure 2-8: SNR-based Partner Selection

If the table is empty at the time of RTS transmission, the address field is left empty and the terminal will be on its own in case of error. This can happen when a terminal just joined the network or when it is too far from the other stations.

Since the terminals estimates the SNR of the packets exchanged on the channel, there is no additional signaling packet or overhead needed to select the partner.

Furthermore, the neighboring terminals are listed by SNR level in the PPT and only the first terminal will be selected to cooperate. This eliminates the overhead added by the contention between the relay candidates.

2.2.3 Coverage Area Extension Evaluation

Current trends in communications are accompanied with an increasing demand in high speed data transmission. On way to achieve high speed communication is to increase the data transmission rate. However as the data rate increases, the signal vulnerability to channel attenuations deepens and the coverage area of a terminal decreases [5]. As a result, a certain data rate can be supported within a limited

(44)

range beyond which a lower rate has to be used. We will show here that using cooperation allows the stations to maintain a certain data rate for a wider coverage area.

We have to evaluate the influence of cooperation on the coverage area of a terminal for a given data rate. To do so, we first define the link margin and find the theoretical coverage area for each data rate. The link margin is the theoretical parameter used to measure how close a link is to failing. It is the difference between the system gains and the system losses as stated in equation (2-12) [40][41].

Link Margin = Transmission Power − Reception Sensitivity

+Antenna Gain − Path Loss (2-12)

The Reception Sensitivity is the minimum amplitude required to receive a signal correctly. It depends on the data rate and the hardware specifications. We carry out our simulations using the specifications of the Cisco Aironet 802.11A/B/G Wireless CardBus Adapter reported in Table 2-2 [42]. We consider the log- distance path loss model with a path loss exponent of 3. The log-distance path loss is given in equation (2-13) and is measured in dB [43].

L = PL(d₀) + N. log₁₀�_d^d

0� + X_S (2-13)

where

PL(d0) is the theoretical path loss at the reference distance, usually taken as free-space loss at 1m,

N is the path loss distance exponent, here it assumed to be 3,

Xs is a Gaussian random variable with zero mean and represents the attenuation caused by fading.

Without loss of generality, we assume that the Antenna Gain = 0dB. A link is safe when the link margin is greater than 10 dB. When the link margin is equal to 0dB, we have the theoretical maximum coverage area.

(45)

Table 2-2: Reception sensitivity, coverage area and network radius for -5dB link margin for different data rates using Cisco Aironet 802.11A/B/G Wireless CardBus Adapter [42]

Data Rate [Mbps] Reception Sensitivity [dBm]

Coverage Area[m]

Network radius with link margin=-5dB[m]

9 -86 75 111

18 -86 75 111

36 -80 47 70

54 -71 24 35

Here, we evaluate the protocol when the link margin is -5dB at the border of the network area in order to find any arbitrary distance at which the signal cannot be successfully decoded anymore. Using equations (2-12) and (2-13), for different data rates, we derive the respective coverage area and the network radius where the link margin would be -5dB at the borders.

2.2.4 Simulation Results

We consider 10 stations randomly placed around an access point. The terminals are deployed within an area at the border of which the Link Margin is - 5dB. This distance varies according to the data rates and the values are found in Table 2-2.

All the stations are mobile with a velocity ranging between 2 and 4 m/s. They move in random directions and therefore, they may move inside or outside the - 5dB link margin borders. They transmit at a constant data rate (9, 18, 36 or 54 Mbps) with the same transmission power and they are in saturated condition.

Figure 2-9 shows the average packet delivery ratio of the network for each “Data rate-Network radius” configuration. We see that cooperation outperforms the non- cooperative protocol in every case. Although the packet delivery ratio decreases as the data rate increases, we can also see that the improvement increases as well.

This is because signals with higher data rates are more error prone and the transmissions are more likely to fail. Consequently the need for cooperation is more significant. In contrast with the non-cooperative protocol, cooperation helps the stations to avoid this failure by retransmitting the data through another

Studies on Cooperative Communication for Wireless LAN

Waseda University Doctoral Dissertation