Throughput and delay limits of chirp spread spectrum-based IEEE 802.15.4a

INTERNATIONAL JOURNAL OF COMMUNICATION SYSTEMSInt. J. Commun. Syst. 2012; 25:1–15Published online 11 May 2011 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/dac.1283

Throughput and delay limits of chirp spread spectrum-basedIEEE 802.15.4a

Niamat Ullah*,†, M. Sanaullah Chowdhury, Pervez Khan, Sana Ullahand Kyung Sup Kwak

Graduate School of Information Technology and Telecommunications, Inha University, Incheon, South Korea

SUMMARY

The IEEE 802.15.4 standard is designed to provide a low-power, low data rate protocol offering a highreliability. As an amendment to this standard, IEEE 802.15.4a introduces new options for physical layerto enable precision ranging. In this work, we analyzed the theoretical throughput and delay bounds of theunslotted version of chirp spread spectrum PHY-based 802.15.4a. The formulae for transmission betweenone sender and one receiver for an ideal channel with no transmission errors are given. The throughput anddelay bounds are derived for different frequency bands and data rates. Additionally, to measure spectral uti-lization, we measured the bandwidth efficiency for both the standards. We also compared our results withIEEE 802.15.4. The comparative analysis concludes that the performance of 802.15.4a exceeds 802.15.4 interms of throughput and delay. The analytical results of throughput are verified by computer simulations.Copyright © 2011 John Wiley & Sons, Ltd.

Received 18 August 2010; Revised 6 March 2011; Accepted 25 March 2011

KEY WORDS: IEEE 802.15.4a; throughput; CSMA/CA; bandwidth efficiency

1. INTRODUCTION

The recent advent in wireless communications triggered the development of standard protocolsspecifically designed for a particular range of applications. As a consequence, the release of IEEE802.15.4 [1] represents a milestone in wireless personal area networks and wireless sensor areanetworks in that the goal of this standard is to provide a low-power, low-cost, and highly reliableprotocol for wireless connectivity among fixed and portable devices [2–4]. Shortly after the releaseof the 802.15.4 standard, it was evident that the range of potential applications of a low bit rate stan-dard could be significantly increased by the capability of measuring the distance between devices inthe network with high accuracy [5]. Because this capability was not included in 802.15.4 devices asa result of the limited signal bandwidth, the IEEE 802.15.4a standard has evolved [6]. The new stan-dard defines alternative physical layers (PHYs) for providing the desired ranging capability and cor-respondingly adapting the medium access control (MAC) layer. The additional supports of rangingin 802.15.4a are expected to enable significant new applications and market opportunities. Sensingand location mapping of disaster sites, precision agriculture, location-based routing, and data collec-tion as well as location tracking of moving objects are some of the applications of IEEE 802.15.4a.

In this paper, we analyze the theoretical throughput and delay bounds for both the CSS-basedIEEE 802.15.4a and the IEEE 802.15.4 for various frequency bands and data rates. We also inves-tigate the bandwidth efficiencies for both the standards. All the information needed to obtain theseresults can be found in the standards [1, 6]. However, it is a laborious procedure that requires data

*Correspondence to: Niamat Ullah, Graduate School of Information Technology and Telecommunications, InhaUniversity, 253 Yonghyun-dong, Nam-gu, Incheon, 402-751, South Korea.

†E-mail: [email protected]

Copyright © 2011 John Wiley & Sons, Ltd.

2 N. ULLAH ET AL.

gathering from various standards and a thorough understanding of mechanisms presented in the stan-dards. This paper offers the exact formulae for these calculations in order to give an overview andan easy way to calculate the maximum throughput (MT), minimum delay, and bandwidth efficiencywithout the need of a complete understanding of the standards.

The rest of the paper is structured as follows. Related works are discussed in Section 2. Section 3introduces brief overviews for both of the standards. In Section 4, the throughput and delay calcu-lations along with the formulae are given. The results and discussions are introduced in Section 5,followed by the conclusion in Section 6.

2. RELATED WORKS

The 802.15.4a PHY is based on ultra wide band (UWB) and chirp spread spectrum (CSS). Theerror-rate performance of IEEE 802.15.4a-compliant UWB radios was investigated in [7]. In [8],the impact of position information on routing was investigated by comparing a traditional solutionbased on the ad hoc on-demand distance vector routing protocol with a position-based solution rely-ing on the combination of the self-positioning algorithm distributed positioning protocol and thegeneral packet radio services routing protocol. Robust and energy-detection receivers for impulseradio–UWB transmission have been presented and analyzed in [9–12]. Taking the impact of thecharacteristics of the new PHY on medium access into account, Luca De Nardis et al. provided anoverview and a comparison of 802.15.4 with 802.15.4a on the MAC layer in [13].

The preliminary performance study of the IEEE 802.15.4a through practical experiments includ-ing numeral wireless nodes, numeral data packets, data transmission with different high-layerprotocols, and physical distance between each node are presented in [14]. The authors designeda wireless network for industrial applications based on 802.15.4a using the CSS technology. In [15],IEEE 802.15.4a CSS-based localization system for wireless sensor network is demonstrated. In [16],the authors made an experiment for the performance of Nexbee, which is one of the CSS-basedlocalization systems. But there exists no paper that encompasses the theoretical limits of 802.15.4abased on the CSS in the literature.

The MT and the minimum delay of 802.15.4 is investigated both analytically and experimentallyin [17]. That paper offers the formulae in order to give an overview and an easy way of calculatingthe MT. The throughput and delay analysis of IEEE 802.15.4 in [18] have considered the unslotted(i.e., without the super frame) version in the 2.4-GHz band. Several papers have addressed theissue of the performance analysis of 802.15.4 which focused on the slotted version in multi-hopenvironment or with multiple senders and receivers. The performance evaluation in [19] studiesthroughput–energy–delay tradeoffs. Many studies [20–23] on IEEE 802.15.4 MAC performancesare concentrated on slotted carrier sense multiple access/collision avoidance (CSMA/CA) by mod-eling the stochastic behavior of one device as a discrete Markov chain. Interesting results in [24]reveal that in a non-beaconed mode and for low-rate applications, the packet delivery ratio of IEEE802.15.4 is similar to IEEE 802.11. According to [25], one of the most common misconceptions isto consider the actual data rate as the throughput. For example, for the data rate of 11 Mbps, theactual throughput offered to a user of IEEE 802.11 for a payload size of 1460 bytes is just 6.1 Mbps(and not 11 Mbps). This is due to the overhead added to the data packet at each layer. The throughputbecomes lower for a higher layer because the overhead accumulates at each layer.

This paper focuses on the unslotted version of 802.15.4 and on the CSS PHY-based 802.15.4athat uses a similar approach as the one used in [26] and [27] for IEEE 802.11.

3. OVERVIEW OF THE STANDARDS

In this section, we give a brief review of the two standards along with the frame sequencing.

3.1. Description of IEEE 802.15.4

The 802.15.4 standard operates in three different frequency bands. It defines 16 channels in the 2.4-GHz band, 10 channels in the 915-MHz band, and 1 channel in the 868-MHz band. All of these

Copyright © 2011 John Wiley & Sons, Ltd. Int. J. Commun. Syst. 2012; 25:1–15DOI: 10.1002/dac

THROUGHPUT AND DELAY LIMITS OF CSS-BASED IEEE 802.15.4A 3

channels use the direct sequence spread spectrum. Typical devices are expected to cover a 10- to 20-m range. The different data rates, modulation techniques, and symbol rates for the correspondingfrequency bands are given in Table I.

The standard defines two channel access modalities:

� beacon-enabled, which uses a slotted CSMA/CA and an exponential backoff (BO), and� nonbeacon-enabled, which uses a simpler unslotted CSMA/CA.

In the beacon-enabled mode, the personal area network coordinator (PNC) broadcasts a periodicbeacon. The time between these beacons is divided into 16 slots of equal sizes. These slots aredivided in two groups, the contention access period and the contention free period. The time slotsin the contention free period are called guaranteed time slots and are assigned by the PNC. Thechannel access in the contention access period is a contention-based CSMA/CA.

The nonbeacon-enabled mode employs the unslotted CSMA/CA channel access mechanism. Theunslotted CSMA/CA operates as follows. When a device needs to send data, it picks a random back-off delay. Subsequently, it checks if the medium is idle. If so, the device transmits the data packet.If the channel is busy, the device repeats the procedure by picking a new backoff delay.

Unlike traditional CSMA/CA with request-to-send/clear-to-send mechanism used in the IEEE802.11, the unslotted CSMA/CA used in IEEE 802.15.4 does not prevent collisions caused byhidden/exposed node problems [7].

It should be noted that although the access protocol is referred to as CSMA/CA, its implemen-tation is closer to a CSMA scheme because no CA packets are used to avoid the collision due to ahidden node problem. This approach differs from wireless local area network IEEE standards, suchas the 802.11 family, where the access protocol foresees the use of request to send and clear to sendbefore sending data packets in order to avoid the collision due to a hidden node problem [7].

3.2. Description of IEEE 802.15.4a

The 802.15.4a standard that addresses the positioning aspects is an evolved version of the parentIEEE 802.15.4. The 802.15.4a PHY is based on two different technologies, UWB and chirp signals.UWB operates in the unlicensed UWB spectrum, and the CSS operates in the unlicensed 2.4-GHzspectrum. We consider only the CSS PHY in this paper.

The chirp solution has the advantage of working in the 2.4-GHz industrial, scientific, and medicalband, thus allowing the deployment of 802.15.4a networks almost worldwide, including in coun-tries where UWB emissions are not yet allowed. For chirp signals, the standard defines 14 channelsspaced at 5 MHz in the frequency range between 2410 and 2486 MHz. The 2450-MHz CSS PHYsupports a data rate of 1 Mbps and optionally, 250 kbps. The two data rates offer flexibility to selectthe rate and properties best suited for different applications. The different data rates, symbol rates,and symbol durations for CSS PHY are given in Table II.

Table I. Parameters of IEEE 802.15.4.

Symbol rate Bit rate Symbol durationFrequency band PHY (symbols/s) Modulation (kbps) (Ts in �s)

868.0–868.6 MHz 868 MHz DSSS 20000 BPSK 20 50902.0–928.0 MHz 915 MHz DSSS 40000 BPSK 40 252.4 –2.4835 GHz 2.4 GHz DSSS 62500 16-ary orth 250 16

DSSS, direct sequence spread spectrum; PHY, physical layer; BPSK, binary phase-shift keying.

Table II. Parameters for chirp spread spectrum physical layer of the IEEE 802.15.4a.

Frequency band PHY Symbol rate Bit rate Symbol duration(GHz) (GHz CSS) (ksymbols/s) (kbps) (Ts in �s)

2.4 –2.4835 2.4 166.667 250 62.4 –2.4835 2.4 166.667 1000 6


4 N. ULLAH ET AL.

Figure 1. Frame structure of 802.15.4a. MAC, medium access control; FCS, frame check sequence; SHR,synchronization header; SFD, start-of-frame delimiter; PHR, PHY header; PSDU, PHY service data unit;

PPDU, PHY protocol data unit.

The complete frame structure of the CSS PHY is illustrated in Figure 1. The payload of the MACprotocol data unit (MPDU) is variable with the limitation that a complete MAC-frame (MPDUor PHY service data unit) may not exceed 127 bytes. The size of the address information variesbetween 0 and 20 bytes as both are short (16 bits), and long addresses (64 bits) can be used. A returnacknowledgment frame does not contain any address information. The address info field can containa 16-bit PAN identifier, both from the sender and from the receiver. These identifiers can be omittedwhen no addresses are sent. Moreover, the preamble for the 1-Mbps data rate consists of eight chirpsymbols, and that of the optional 250 kbps consists of a higher number of chirp symbols, equal to20. For the two different data rates, the CSS PHY also specifies different start-of-frame delimiterbit sequences. The MAC strategies described in Section 3.1 of the IEEE 802.15.4 are inherited inIEEE 802.15.4a with a significant difference in the channel access mechanism. ALOHA is intro-duced as an alternative channel access strategy. This decision is based on the multi-user interferencerobustness guaranteed by the UWB PHY that enables the ALOHA approach to provide satisfactorythroughput in medium and lightly loaded networks. The CSMA/CA access is kept as an option inorder to address high-density and high-traffic scenarios and to enable the use of the CSS PHY. To thisextent, the CSMA/CA access scheme would likely find applications in areas such as home automa-tion and consumer electronics, health-care utilities, traffic regulation, and military applications suchas battlefield surveillance and army deployment.

3.3. Frame Sequence

As the MAC sub layer requires a finite amount of time to process the data received from the PHY,the transmitted frames are followed by an interframe space (IFS) period. Long and short frameswill be followed by a long IFS (LIFS) and a short IFS (SIFS), respectively. An example of a framesequence, used with and without acknowledgments (ACKs) is given in Figure 2. If no ACKs areused, the IFS follow the frame immediately.

4. THROUGHPUT AND DELAY CALCULATIONS

Because the 802.15.4/4a standards cover the MAC and the PHY in terms of the Open SystemInterconnection reference model [18], we are interested in the actual throughput provided by the



(c)

(b)

(a)

Figure 2. (a) Timing diagram of carrier sense multiple access/collision avoidance (CSMA/CA) with inter-frame space (IFS) mechanisms, and the frame sequence (b) with and (c) without acknowledgments. ACK,acknowledgment; SIFS, short IFS; LIFS, long IFS; TA, turnaround; BO, backoff; MPDU, medium access

control protocol data unit.

MAC layer. Therefore, we define the MT as the maximum number of MAC layer service data units(MSDUs or payload) that can be transmitted in a unit of time. Each MSDU or payload carries anadditional overhead at the MAC layer and the PHY such as MAC/PHY header, address info, MACfooter, BO time, TA time, ACK transmission time, and IFS time. This additional overhead affectsthe MT in the network. We derive numerical approximations and formulae valid for different fre-quency bands and data rates. We only examine the unslotted version of both the standards. As thisversion has the least overhead, it will produce an upper bound on the MT of the protocols.

The MT is calculated between only one sender and one receiver, which are located close to eachother. In addition, the following assumptions are considered for the calculations: (i) BER is zero;(ii) there are no losses due to collisions; (iii) there is no packet loss due to the buffer overflow at thereceiving node; and (iv) the sending node always has sufficient packets to send.

The delay is the time needed to transmit one packet. The overall delay accounts for the delay ofdata being sent and for the delay caused by all elements as shown in Figure 2. Each packet delaycan be expressed as:

Total delay D Durationframe(pld)CDurationBO, TA, ACK, IFS

Delaytotal D tframe(pld)C tBOC tTAC tACKC tIFS , (1)

where tframe(pld) is the transmission time for a payload, pld, bytes that has been received from anupper layer, tBO is the BO period in seconds, tTA is the TA time, tACK is the ACK transmission time,and tIFS is the IFS time.

The total duration of the frame is:

tframe(pld) D 8 � ŒLPHY_OvrhdC ¹LMAC_HdrCLMAC_FtrC .LaddrC pld/º�=RateData , (2)


6 N. ULLAH ET AL.

where LPHY_Ovrhd is the length of PHY overhead in bytes, LMAC_Hdr is the length of MAC header inbytes, LMAC_Ftr is the length of MAC footer in bytes, Laddr is the length of MAC address info fieldin bytes, and RateData is the raw data rate.

The values for the above parameters can be found in Table III.The Laddr is the total length of MAC address including the PAN identifiers for both the sender

and the receiver. The length of a PAN identifier is 2 bytes.The backoff period is expressed as:

tBO D average number of backoff slots � time for a backoff slot. (3)

The number of backoff slots is a random number uniformly distributed in (0, 2BE � 1) with abackoff exponent (BE) that has a default value of 3. As we only assume one sender and a BER ofzero, the BE will not change. Hence, the number of backoff slots can be represented as the mean ofthe interval: .2BE � 1/=2 or 3.5.

The timing parameters can be found as:

tslots D 20.Ts , tTA D 12.Ts , tSIFS D 12.Ts , tLIFS D 40.Ts , (4)

where Ts represents the duration of one symbol. The values of Ts can be derived and found fromTables I and II for different frequency bands for both the standards.

Long IFS is used when the MPDU size is greater than 18 bytes; otherwise, SIFS is used.The duration of ACKs can be calculated as:

tACK D 8 � ¹LPHY_OvrhdC .LMAC_HdrCLMAC_Ftr/º=RateData (5)

If no ACKs are used, tTA and tACK are omitted.Finally, we can express the delay and throughput using the following formulae:

Delay.pld/D P � .pld/CQ (6)

ThroughputD .8 � pld/=Delay.pld/ (7)

The parameter P expresses the delay needed for sending one data byte, and the parameter Q is thetime needed for the protocol overhead to send one packet. Different values of P and Q are shownin Table IV. The parameters P and Q depend on the length of the data bytes (SIFS or LIFS), thelength of the address used (64 bit, 16 bit, or no address), and the frequency bands.

The bandwidth efficiency is the ratio of throughput to the desired data rate and can beexpressed as:

¡D Throughput=RateData (8)

The maximum size of MPDU, which is 127 bytes, shows the number of data bytes that can be sentin one packet is limited.

Table III. Parameters of IEEE 802.15.4 and IEEE 802.15.4a.

Parameters 802.15.4 802.15.4a

RateData 250 kbps 250 kbps 1 MbpsLPHY_Ovrhd 48 bits 128 bits 56 bitsLMAC_Hdr 24 bits 24 bits 24 bitsLMAC_Ftr 16 bits 16 bits 16 bitsLaddr 0 –160 bits 0 –160 bits 0 –160 bitstTA 192 �s 72 �s 72 �stLIFS 640 �s 240 �s 240 �stSIFS 192 �s 72 �s 72 �stslots 320 �s 120 �s 120 �s



Table IV. Values of P and Q in Equation (6).

IEEE 802.15.4 (2.4 GHz) IEEE 802.15.4a (CSS PHY)Nominal data rate Nominal data rate Optional data rate

(250 kbps) (1 Mbps) (250 kbps)No. of address bits P Q P Q P Q

0 No ACK 0.000032 0.002112 0.000008 0.000756 0.000032 0.0013320 ACK 0.000032 0.002656 0.000008 0.000924 0.000032 0.00207616 No ACK 0.000032 0.002368 0.000008 0.00082 0.000032 0.00158816 ACK 0.000032 0.002912 0.000008 0.000988 0.000032 0.00233264 No ACK 0.000032 0.002752 0.000008 0.000916 0.000032 0.00197264 ACK 0.000032 0.003296 0.000008 0.001084 0.000032 0.002716

ACK, acknowledgment; CSS, chirp spread signal; PHY, physical layer.

Table V. Maximum throughput (bps) and bandwidth efficiency (%).

IEEE 802.15.4 (2.4 GHz) IEEE 802.15.4a (CSS PHY)(250 kbps) (1 Mbps) (250 kbps)

No. of Throughput Bandwidth Throughput Bandwidth Throughput Bandwidthaddress bits efficiency efficiency efficiency

0 No ACK 162234 64.89 563510 56.35 186402 74.560 ACK 148780 59.51 513684 51.37 163211 65.2816 No ACK 151596 60.64 526559 52.66 174179 69.6716 ACK 139024 55.61 480000 48 152509 61.0064 No ACK 135638 54.26 471132 47.11 155844 62.3464 ACK 124390 49.76 429474 42.95 136455 54.59


In the case of a 0-bit address, the payload size can be at most 122 bytes. In the case of a shortaddress (2 bytes for each sender and receiver and 2 bytes for each PAN identifier), the payload sizeis 114 bytes. For a long address (8 bytes for each sender and receiver and 2 bytes for each PANidentifier), the payload size is 102 bytes. Generally the MPDU size can be calculated as:

MPDUsizeD LMAC_HdrCLMAC_FtrCLaddrC pld (9)

Using the equations mentioned, the values of the throughput and bandwidth efficiency of 802.15.4and 802.15.4a are found and put into Table V.

5. RESULTS AND DISCUSSION

In this section, we present the throughput, the delay, and the bandwidth efficiency of both standardsfor different data rates. In our analysis, we considered the 250 kbps and 1 Mbps for the 802.15.4astandard and the 250 kbps for the 802.15.4 standard. The parameters for both the standards are givenin Table III. The values of P andQ for different data rates of the corresponding standards are foundin Table IV.

Figures 3 and 4 show the minimum delay and the MT as functions of payload size, respectively.It can be seen that the throughput is quite low for the optional data rate of 802.15.4a and for thenominal data rate of 802.15.4.

Table V shows that when the payload size is 122 bytes with no ACK policy, the MT for 1 Mbps of802.15.4a is 563.5 kbps, the MT for 250 kbps of 802.15.4a is 186.4 kbps, and the MT for 802.15.4is 162.2 kbps. For the payload size of 102 bytes (i.e., 64-bit address) with ACK policy, the MT for802.15.4a with data rate 1 Mbps is 429.5 kbps, the MT for 802.15.4a with data rate 250 kbps is136.5 kbps, and the MT for 802.15.4 is 124.4 kbps.

Table VI gives the minimum and maximum delay for different scenarios for both standards. Theminimum delay is calculated by taking a 0-bit payload. We do not consider any retransmission or


8 N. ULLAH ET AL.

0 20 40 60 80 100 1200

1

2

3

4

5

6

7

8

Payload size (bytes)

Del

ay (

ms)

802.15.4 (Adr(0), 250 kbps)CSS (Adr(0), 1 Mbps)

CSS (Adr(0), 250 kbps)

802.15.4 (Adr(64), ACK, 250 kbps)

CSS (Adr(64), ACK, 1 Mbps)CSS (Adr(64), ACK, 250 kbps)

Figure 3. Comparison of the minimum delay as a function of payload size. CSS, chirp spread spectrum;ACK, acknowledgment.

0 20 40 60 80 100 1200

1

2

3

4

5

6x 10


Th

rou

gh

pu

t (b

ps)

802.15.4 (Adr(0), no ACK, 250 kbps)CSS (Adr(0), no ACK, 1 Mbps)

CSS (Adr(0), no ACK, 250 kbps)

802.15.4 (Adr(64), ACK, 250 kbps)

CSS (Adr(64), ACK, 1 Mbps)CSS (Adr(64), ACK, 250 kbps)

5

Figure 4. Throughput comparison of 802.15.4 and 802.15.4a for different address schemes. CSS, chirpspread spectrum; ACK, acknowledgment.

Table VI. Maximum and minimum delay (ms).

IEEE 802.15.4 (2.4GHz) IEEE 802.15.4a (CSS PHY)(250 kbps) (1 Mbps) (250 kbps)

No. of address bits Min delay Max delay Min delay Max delay Min delay Max delay

0 No ACK 1.664 6.016 0.588 1.732 1.164 5.2360 ACK 2.208 6.56 0.756 1.9 1.911 5.98316 No ACK 1.92 6.016 0.652 1.732 1.42 5.23616 ACK 2.464 6.56 0.82 1.9 2.167 5.98364 No ACK 2.304 6.016 0.748 1.732 1.804 5.23664 ACK 2.848 6.56 0.916 1.9 2.551 5.983




propagation delay. Minimum delay means the time needed for sending an empty packet from onesender to one receiver. In this table, we calculate the minimum delay for various scenarios like 0-,2-, and 8-byte addresses with and without ACKs.

The maximum delay is found by sending a frame by considering a maximum packet size, that is,the MPDU is set to 127 bytes. The maximum payload size differs by changing the address schemes.In the case of a 0-byte address scheme, the maximum payload can be 122 bytes. Similarly fora 2-byte address scheme, the maximum payload is 114 bytes, and the maximum payload can be102 bytes for an 8-byte address scheme.

The minimum delay for different payload sizes in the 2.4-GHz band for both 802.15.4 and802.15.4a can be seen in Figure 3. Here, we can see that the delay is much higher for lower data rates.We can see a delay of up to 6.56 ms in the case of 802.15.4. However, this delay is an acceptabledelay bound application. The delay is well reduced in 802.15.4a with a data rate of 1 Mbps. Thesedelay figures and tables can offer a more thorough insight on the limitations of both the standardswhen designing and implementing a network based on these protocols.

Figure 3 also shows that in the case of the same data rate for both 802.15.4a and 802.15.4, thedelay of 802.15.4a is smaller than that of 802.15.4 for both short and long addresses. In the case ofthe different data rates of 802.15.4a, the delay for 1 Mbps is smaller. The reason behind this smallerdelay is due to the smaller symbol duration and data rate. For a payload size of 102 bytes with ACKpolicy, the minimum delay for 802.15.4a with a data rate of 1Mbps is around 1.65 ms, the minimumdelay for 802.15.4a with a data rate of 250 kbps is approximately 5.2 ms, and the minimum delayfor 802.15.4 with a data rate of 250 kbps is close to 6 ms.

Figure 4 shows the comparison of throughput obtained when different address schemes are used.We see that the curve of IEEE 802.15.4a having a 1 Mbps data rate with 0-bit address and no ACK ishigher than the others. The reason for this higher throughput is due to a less control traffic, a higherdata rate, and a smaller symbol duration. In Figure 5, the payload size represents the number ofbytes coming from the upper layer. Generally the throughput increases with the increase of payload.

In Figure 5, the throughput of IEEE 802.15.4a with the same data rate as that of IEEE 802.15.4is higher because IEEE 802.15.4a uses smaller symbol duration. In addition, the MT for low datarates saturates much earlier than that of the higher data rates. Figures 4 and 5 present the theoreticalthroughput and delay bounds of both the standards.

Figure 5 describes the small jumps in the graphs. These jumps are due to the facts that SIFS areused for MPDUs of up to 18 bytes in length and that LIFS are used otherwise. The jump occurswhen the MPDU size becomes greater than 18 bytes.

2 4 6 8 10 12 14 16 18 200

2

4

6

8

10

12

14

16

18x 104


Th

rou

gh

pu

t (b

ps)

802.15.4 (Adr(0), 250 kbps)

CSS (Adr(0), 1 Mbps)

CSS (Adr(0), 250 kbps)802.15.4 (Adr(16), 250 kbps)

CSS (Adr(16), 1 Mbps)

Figure 5. The transition from short interframe space to long interframe space. CSS, chirp spread spectrum.


10 N. ULLAH ET AL.

In Figure 5, we can see that in case of a no-address scheme, the fluctuation occurs when thepayload size becomes greater than 13 bytes. In the case of a 16-bit address, we can see this changewhen the payload size becomes greater than 6 bytes. In Figures 3, 4, and 6, we cannot see any kindof jump for a 64-bit address scheme. The reason is that we cannot use SIFS, even for a payload of0 bit; instead, we use LIFS.

We now present the bandwidth efficiency, � in order to measure the spectral utilization of boththe standards. The � is inversely proportional to the basic data rate. It can be seen in Figure 6 that� is higher for low data rates and increases with the increase in payload size. The throughput andbandwidth efficiency results for varying types of payload sizes and different address schemes arebriefly summarized in Table V.

We can see the bandwidth efficiency for a payload size of 122 bytes for 802.15.4a with 250 kbpsand 1 Mbps data rates are 74.56% and 56.35%, respectively, and for 802.15.4 is 64.89%.

The logic behind this higher bandwidth efficiency is the use of optimal circumstances, that is,when no address and no ACKs are used. If ACKs are used, an efficiency of only 65.28% is obtained.Using short addresses, that is, a 2-byte address for each sender and receiver, further lowers band-width efficiency and throughput. The worst result is an efficiency of only 42.95%, which is reachedin the case of using addresses 8 bytes in length for each sender and receiver with ACKs. The reasonfor low bandwidth efficiency is that of relatively more overhead bytes as compared with an actualpayload. In Figure 6, we have only shown the graphs for a no-address scheme (0-bit address) and along-address scheme (64-bit address) with ACK because these two are the extreme cases.

In the case of a short-address scheme, a similar conclusion can be drawn. Figure 7 shows acomparison of bandwidth efficiency of the different rates of 802.15.4a standard for a short-addressscheme (16-bit). From the graph, it is clear that the bandwidth efficiency increases when no ACKis used, which is to be expected as less control traffic is being sent. A summary can be found inTable V where the MT and bandwidth efficiency of the scenarios are given. We see that bandwidthefficiency is lower for high data rates. The reason for these low results is that the length of MPDUis limited to 127 bytes.

Figures 8 and 9 show the influence of different MAC minimum BE values on throughput andbandwidth efficiency. From those figures, we found that in the case of a single sender and a singlereceiver, the throughput and bandwidth efficiency can be improved by taking a small backoffinterval. For example, in the case of the 1-Mbps CSS PHY (when a 16-bit address with ACK isused), the efficiency decreases from 62% (when BE D 0) to 28% (when BE D 5). However, in thecase of multiple senders, this small interval will result in collisions and will influence the networkthroughput as well as the bandwidth efficiency significantly.

0 20 40 60 80 100 1200

10

20

30

40

50

60

70

80


Ban

dw

idth

eff

icie

ncy

(%

)

802.15.4 (Adr(0bit), 250 kbps)

802.15.4 (Adr(64bit), 250 kbps)

CSS (Adr(0bit), 1 Mbps)CSS (Adr(64bit), 1 Mbps)

CSS (Adr(0bit), 250 kbps)

CSS (Adr(64bit), 250 kbps)

Figure 6. Bandwidth efficiency for a varying payload size.



0 20 40 60 80 100 1200

10

20

30

40

50

60

70

80


Ban

dwid

th e

ffici

ency

(%)

CSS (ACK, 1 Mbps)

CSS (ACK, 250 kbps)

CSS (noACK, 1 Mbps)

CSS (noACK, 250 kbps)

Figure 7. Comparison of bandwidth efficiency of chirp spread spectrum (CSS)-based 802.15.4a using shortaddresses. ACK, acknowledgment.

0 1 2 3 4 5

1

2

3

4

5

6

7

8x 105

MACminBE

Th

rou

gh

pu

t (b

ps)

CSS (250 kbps, noACK, Adr(0))

CSS (1 Mbps, Adr(0), noACK)

CSS (250 kbps, Adr(64), ACK)

CSS (1 Mbps, Adr(64), ACK)

Figure 8. Useful bit rate for varying medium access control minimum backoff exponent (MACminBE)values. CSS, chirp spread spectrum; ACK, acknowledgment.

We validate the analytical results through extensive simulations using the NS-2 (version 2.31).We consider a total of three nodes including the PNC in a star topology. The nodes generate Pois-son traffic in the uplink direction. The simulation area is 3 � 3 m. The parameters are consideredaccording to Table III. The shadowing propagation model is used throughout the simulations. Theinitial node’s energy is 5 Joules. The average packet size is 60 bytes. The data rates for CSS-basedIEEE 802.15.4a are 1 Mbps and 250 kbps. The User Datagram Protocol is used as a transport agent.The buffer size at the coordinator is unlimited. Because we are interested to calculate the through-put and delay limits at the MAC layer, the no ad-hoc (NOAH) is used as a routing agent. The totalsimulation time is 500 s. For analytical analysis, we have used only one sender and one receiver, butin simulation, the use of one sender and one receiver is meaningless; therefore, we used three nodes.Each node generates a data traffic directed towards the PNC, but the PNC only generates ACK. Weestimate the throughput by measuring the total number of bytes received at the PNC and the delay


12 N. ULLAH ET AL.

0 1 2 3 4 50

10

20

30

40

50

60

70

80

MACminBE

Ban

dw

idth

eff

icie

ncy

(%

)802.15.4 (250 kbps, noACK)

CSS (1 Mbps, noACK)

CSS (250 kbps, noACK)802.15.4 (250 kbps, ACK)

CSS (1 Mbps, ACK)

CSS (250 kbps, ACK)

Figure 9. Bandwidth efficiency for varying medium access control minimum backoff exponent(MACminBE) values with short-address scheme. CSS, chirp spread spectrum; ACK, acknowledgment.

0 20 40 60 80 100 1200

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5x 105

Payload (bytes)

Thr

ough

put

(bp

s)

Anal (Adr(64), 1 Mbps)

Sim (Adr(64), 1 Mbps)Anal (Adr(16), 1 Mbps)

Sim (Adr(16), 1 Mbps)

Anal (Adr(64), 250 kbps)

Sim (Adr(64), 250 kbps)Anal (Adr(16), 250 kbps)

Sim (Adr(16), 250 kbps)

Figure 10. Comparison between simulation and analytical results.

by measuring the time interval when a packet is generated until it is successfully acknowledged.The final results are plotted using MATLAB (Matrix Laboratory (Mathworks, Inc.)).

Figure 10 gives a throughput comparison between the analytical and the simulation results inthe cases of short and long addresses with ACK schemes. According to the expectation that thetheoretical analysis offers an upper bound to the throughput, we see that the simulation curves arelower than analytical curves. This difference is mainly due to the delays caused by the occurrenceof retransmissions due to collisions.

6. CONCLUSION

In this paper, we presented the theoretical MT and minimum delay bounds along with bandwidthefficiency for both the CSS-based IEEE 802.15.4a and the IEEE 802.15.4. The MT and the minimumdelay were determined by considering communications between one sending radio and one receiv-ing radio. We have considered the unslotted version of 802.15.4 and the CSS PHY of 802.15.4a for



analyzing the performance of MAC layer for both standards. We presented formulae for determin-ing the throughput and the delay for different scenarios. The comparative analysis concludes thatthe performance of 802.15.4a surpasses that of the 802.15.4 in terms of throughput and delay. In802.15.4a, the bandwidth efficiency of 74.56% can be achieved with a data rate of 250 kbps.

On the other hand, an MT of 563 kbps can be achieved with a data rate of 1 Mbps for the samestandard. From the results, we conclude that efficiency can be improved by increasing the payloadsize. We hope that the results of this paper will help researchers and system designers to correctlyprovision systems based on the CSS-based IEEE 802.15.4a technology.

In future, we are interested to consider more complex and realistic analytical models in order topredict the real performance of the network. The analytical results will be verified using the (NS)simulations under different scenarios. Also, we will analyze the performance of slotted ALOHA forUWB-based IEEE 802.15.4a networks.

ACKNOWLEDGEMENTS

This research was supported by the MKE (The Ministry of Knowledge Economy), Korea, under the ITRC(Information Technology Research Center) support program supervised by the NIPA (National IT IndustryPromotion Agency) (NIPA-2011-C1090-1121-0001) and by the National Research Foundation of Korea(NRF) grant funded by the Korea government (MEST)(No. 2010-0018116).

REFERENCES

1. IEEE Std 802.15.4-2006, Part 15.4: Wireless Medium Access Control (MAC) and Physical Layer (PHY) Specifica-tions for Low-Rate Wireless Personal Area Networks (LR-WPANs), 2006.

2. Callaway E, Gorday P, Hester L, Gutierrez JA, Naeve M, Heile B, Bahl V. Home networking with IEEE802.15.4: a developing standard for low-rate wireless personal area networks. IEEE Communications Magazine2002; 40(8):70 –77.

3. Zheng J, Lee MJ. Will IEEE 802.15.4 make ubiquitous networking a reality?: a discussion on a potential low power,low bit rate standard. IEEE Communications Magazine 2004; 42(6):140 –146.

4. Gutierrez JA, Naeve M, Callaway E, Bourgeois M, Mitter V, Heile B. IEEE 802.15.4: a developing standard forlow-power low-cost wireless personal area networks. IEEE Network 2001; 15(5):12 –19.

5. Karapistoli E, Pavlidou F-N, Tsetsinas I. An overview of the IEEE 802.15.4a standard. IEEE CommunicationsMagazine 2010; 48(1):47 –53.

6. IEEE Std 802.15.4a-2007, Amendment to 802.15.4-2006: Wireless Medium Access Control (MAC) and PhysicalLayer (PHY) specifications for Low-Rate Wireless Personal Area Networks (LR-WPANs), 2007.

7. Ahmadian Z, Lampe L. Performance analysis of the IEEE 802.15.4a UWB system. IEEE Transactions onCommunications 2009; 57(5):1474 –1485.

8. De Nardis L, Domenicali D, Di Benedetto M-G. Performance and energy efficiency of position-based routing inIEEE 802.15.4a low data rate wireless personal data networks. Proceedings of the IEEE International Conference onUltra-Wideband 2007 (ICUWB 2007), Singapore, 24 –26 September 2007; 264 –269.

9. Choi JD, Stark WE. Performance of ultra-wideband communications with suboptimal receivers in multipathchannels. IEEE Journal on Selected Areas in Communications 2002; 20(9):1754 –1766.

10. Weisenhorn M, Hirt W. Robust noncoherent receiver exploiting UWB channel properties. Proceedings of the2004 International Workshop on Ultra Wideband Systems Joint with Conference on Ultra Wideband Systems andTechnology (Joint UWBST & IWUWBS 2004), Kyoto, Japan, 18 –21 May 2004; 156 –160.

11. Tian Z, Sadler BM. Weighted energy detection of ultra-wideband signals. Proceedings of the 2005 IEEE 6th Work-shop on Signal Processing Advances in Wireless Communications (SPAWC 2005), New York, NY, USA, 6 – 8 June2005; 158–162.

12. D’Amico AA, Mengali U, Arias-de-Reyna E. Energy-detection UWB receivers with multiple energy measurements.IEEE Transactions on Wireless Communications 2007; 6(7):2652 – 2659.

13. De Nardis L, Di Benedetto MG. Overview of the IEEE 802.15.4/4a standards for low data rate wireless personal datanetworks. Proceedings of the 4th Workshop on Positioning, Navigation and Communication, 2007 (WPNC 2007),Hannover, Germany, 22 March 2007; 285–289.

14. Li T, Fei M, Hu H. Performance analysis of industrial wireless network based on IEEE 802.15.4a. In Life SystemModeling and Intelligent Computing. Springer: Berlin Heidelberg, 2010; 64 – 69.

15. Kim J-E, Jihoon K, Kim D, Ko Y, Kim J. IEEE 802.15.4a CSS-based localization system for wireless sensor net-works. Proceedings of the IEEE International Conference on Mobile Adhoc and Sensor Systems, 2007 (MASS 2007),Pisa, Italy, 8 –11 October 2007; 1–3.

16. Kang H, Seo GW, Lee J. Error compensation for CSS-based localization system. Proceedings of the World Congresson Engineering and Computer Science 2009 (WCECS 2009), San Francisco, CA, USA, 20 – 22 October 2009;696 –701.


14 N. ULLAH ET AL.

17. Latré B, De Mil P, Moerman I, VD N, Dhoedt B, Demeester P. Maximum throughput and minimum delay in IEEE802.15.4. In Lecture Notes on Computer Science 3794. Springer-Verlag: Berlin Heidelberg, 2005; 866 – 876.

18. Latré B, De Mil P, Moerman I, Dhoedt B, Demeester P. Throughput and delay analysis of unslotted IEEE 802.15.4.Journal of Networks May 2006; 1(1):20 – 28.

19. Lu G, Krishnamachari B, Raghavendra C. Performance evaluation of the IEEE 802.15.4 MAC for low-rate low-powerwireless networks, Phoenix, Arizona, USA, 15–17 April 2004; 701–706.

20. Misic J, Misic VB. Queueing analysis of 802.15.4 beacon enabled PAN. In Proceedings of the IEEE/ACM FirstInternational Workshop on Broadband Wireless Services and Applications BroadWISE 2004, San Jose, CA, USA,25–29 October 2004.

21. Pollin S, Ergen M, Ergen SC, Bougard B, Van der Perre L, Catthoor F, Moerman I, Bahai A, Varaiya P. Performanceanalysis of slotted IEEE 802.15.4 medium access layer. Technical Report, Interuniversity Microelectronics Center,Belgium, 2005. draft-jwl-tcp-fast-01.txt.

22. Park TR, Kim TH, Choi JY, Choi S, Kwon WH. Throughput and energy consumption analysis of IEEE 802.15.4slotted CSMA/CA. Electronics Letters 2005; 41(18):1017–1019.

23. Kim TO, Kim H, Lee J, Park JS, Choi BD. Performance analysis of IEEE 802.15.4 with non-beacon-enabledCSMA/CA in non-saturated condition. Proceedings of the International Conference of Embedded and UbiquitousComputing 2006 (EUC 2006), Lecture Notes in Computer Science 4096, Seoul, South Korea, 1; 884 – 893.

24. Zheng J, Lee MJ. A Comprehensive Performance Study of IEEE 802.15.4. In Sensor Network Operations. IEEEPress: Los Alamitos, CA, 2006; 218–237.

25. Xiao Y, Rosdahl J. Throughput and delay limits of IEEE 802.11. IEEE Communications Letter 2002; 6(8):355 – 357.26. Jun J, Peddabachagari P, Sichitiu M. Theoretical maximum throughput of IEEE 802.11 and its applications,

Cambridge, MA, USA, 16 –18 April 2003; 249–256.27. Tanenbaum AS. Computer Networks (4th edn.) Prentice-Hall: Upper Saddle River, NJ, 2002.28. The Network Simulator-ns-2. Available from: http://www.isi.edu/nsnam/ns/.

AUTHORS’ BIOGRAPHIES

Niamat Ullah received his B.Sc. in Mathematics in 1994 from Peshawar Universityand Masters in Computer Sciences in 1996 from Quaid-e-Azam University Islamabad,Pakistan. He then joined the Higher Education Department Khyber Pakhtunkhwa as a lec-turer. Currently, he is a PhD student at the Graduate School of IT and Telecommunicationsat Inha University Incheon, South Korea. His research interests include wireless ad hocnetworks, directional and smart antennas, wireless sensor networks and MAC protocol forWireless Body Area Networks.

M. Sanaullah Chowdhury received his Master in Information and Communication Engi-neering from Inha University, South Korea in 2009. Currently he is a PhD studentin the same University. He received his B.Sc. in Computer Science & Engineeringfrom International Islamic University Chittagong, Bangladesh. His research interests areWireless Networking, MAC protocol for Wireless Body Area Networks and Cognitivenetworks.

Pervez Khan has received his Master and Bachelor degrees, both in Computer Sciencefrom the University of Peshawar in 2006 and 2003 respectively, and he is currently workingtowards the Ph.D. degree at the Graduate School of IT and Telecommunication Engineer-ing in Inha University, South Korea. His research interests are wireless communications,wireless sensor networks, MAC protocol for Wireless Body Area Networks and seamlesshandover schemes for heterogeneous wireless networks.



Sana Ullah is a Ph.D. student at the Graduate School of IT & Telecommunications at InhaUniversity Incheon, South Korea. He received his Master of Science in Computer Scienceand his Bachelor of Science in Mathematics, both from the University of Peshawar in2006 and 2003 respectively. His research interest includes MAC protocols for WBAN,low-power implant communication, and cross-layer design. He is a member of IEEE TG6,International Association of Engineers (IAENG), and Finnish Society of Telemedicine andeHealth.

Kyung Sup Kwak received the B.S. degree from Inha University, Korea in 1977, and theM.S. degree from the University of Southern California in 1981 and the Ph.D. degree fromthe University of California at San Diego in 1988, respectively. From 1988 to 1989 he wasa Member of Technical Staff at Hughes Network Systems, San Diego, California. From1989 to 1990 he was with the IBM Network Analysis Center at Research Triangle Park,North Carolina. Since then he has been with the School of Information and Communica-tion, Inha University, Korea as a professor. He had been the chairman of the School ofElectrical and Computer Engineering from 1999 to 2000 and the dean of the GraduateSchool of Information Technology and Telecommunications from 2001 to 2002 at InhaUniversity, Inchon, Korea. He is the current director of Advanced IT Research Center ofInha University, and UWB Wireless Communications Research Center, a key governmentIT research center, Korea. He has been the Korean Institute of Communication Sciences

(KICS)’s president of 2006 year term. In 1993, he received Engineering College Young Investigator AchievementAward from Inha University, and a distinguished service medal from the Institute of Electronics Engineers ofKorea (IEEK). In 1996 and 1999, he received Distinguished Service medals from the KICS. He received the InhaUniversity Engineering Paper Award and the LG Paper Award in 1998, and Motorola Paper Award in 2000. Hisresearch interests include multiple access communication systems, mobile communication systems, UWB radiosystems and ad-hoc networks, high-performance wireless Internet.


Throughput and delay limits of chirp spread spectrum-based IEEE 802.15.4a

Documents

Transcript of Throughput and delay limits of chirp spread spectrum-based IEEE 802.15.4a