Multi Data Rate Signaling based on IEEE 802.15.4

Sebastian Rickers, Christoph Spiegel, Guido Bruck, Peter Jung

Department of Communication Technologies University of Duisburg-Essen

Duisburg, Germany

Rami Lee, SeongSoo Park, Jaehwang Yu Convergence Technology Center

Ambient Technology Lab SK Telecom Seoul, Korea

Abstract—The IEEE 802.15.4 standard was brought up for the use in wireless personal area networks (WPANs) and is widely known as ZigBee. It employs spread spectrum techniques to obtain a robust transmission at comparatively low data rates of up to 250 kbit/s. In order to better match a certain transmission environment, different spreading factors could be used to obtain additional data rate modes. This approach appears to be a feasible extension to IEEE 802.15.4 allowing the coverage of a wider range of the signal to interference and noise ratio (SINR). However, a signaling scheme must be added to distinguish the additional modes from the native ZigBee mode. This scheme should also allow the operation of extended devices without interfering native ZigBee devices. Based on the structure of an IEEE 802.15.4 frame a method is presented that will allow extending the specified frame structure to a multi data rate system without interfering IEEE 802.15.4 compliant devices.

Index Terms—IEEE 802.15.4, ZigBee, wireless personal area network (WPAN), multi data rate extension, start-of-frame delimiter (SFD), 2.45 GHz ISM (industrial, scientific, medical) band

I. INTRODUCTION The ZigBee Alliance is an association of companies that

creates a software stack based on the IEEE 802.15.4 standard. This is the underlying standard defining the medium access control (MAC) sublayer and physical (PHY) layer with a set of transmission parameters in several unlicensed frequency bands, where the fastest transmission mode uses a data rate of 250 kbit/s. Depending on the environment where the system is used it might be useful to either increase or decrease the data rate. On the one hand, in environments with high receive power or low interference it appears to be useful to offer higher data rates to the devices in order to shorten the transmit time. On the other hand, in unfavorable scenarios with low receive power or high interference, a more robust mode could be offered to still be able to transmit data. The latter case would correspond to an increased coverage compared to native ZigBee devices and can be achieved by increasing the spreading factor.

Based on this idea, multi data rate signaling will be presented that will not change the frame structure of IEEE 802.15.4. The reuse of IEEE 802.15.4 compliant hard- and software where just minor changes have to be made is one of the major considerations of the presented scheme. A signaling technique will be presented that extends the start-of-frame delimiter (SFD) while not modifying the IEEE 802.15.4

compliant preamble. The user data following the SFD field will be prepared in a way which allows reusing the native ZigBee receiver structure.

This paper is structured as follows. Section II covers the definition of the physical transmission of IEEE 802.15.4 compliant signals. Section III then explains the technique of finding an optimal set of additional SFD sequences for indicating additional data rates, while Section IV presents simulation results showing the performance of an extended system employing six additional data rates. Section V finally concludes the manuscript.

II. THE IEEE 802.15.4 SPECIFICATION The IEEE 802.15.4 specification defines the characteristics

of the PHY layer and the MAC sublayer. While the MAC sublayer manages the access to the communication medium, carries out the flow control and validates received frames, the PHY layer constitutes the interface to the physical medium.

First of all, the so-called PHY protocol data unit (PPDU) is assembled as shown in Fig. 1. An IEEE 802.15.4 compliant frame contains a preamble of 4 bytes in length, a 1 byte start-of-frame delimiter (SFD), a 1 byte long payload length field and the actual user data with a maximum length of 127 bytes provided by the MAC sublayer. Preamble and SFD field are named synchronization header (SHR). The payload length field is also referred to as the PHY header (PHR) and the user data is termed PSDU (PHY service data unit).

SHR PHR PHY payload

Preamble SFD Frame length (7 bits)

Reserved (1 bit) PSDU

4 bytes 1 byte 1 byte variable length

Figure 1. Fields of an IEEE 802.15.4 PPDU (from [1]).

In addition to this frame structure, modulation and a pulse shaping are defined as well. IEEE 802.15.4 defines a transmission in the 868 MHz band in Europe, in the 915 MHz band in North America and in the global 2.45 GHz ISM (industrial, scientific and medical) band. The supported data rates range from 20 kbit/s up to 250 kbit/s with different modulation techniques such as amplitude shift keying (ASK), binary phase shift keying (BPSK) and offset quaternary phase

978-1-4244-8132-3/10/$26.00 ©2010 IEEE

shift keying (O-QPSK). All these modulation techniques map one bit to one symbol, they can therefore be considered simple and robust. The specified combinations of modulation schemes and data rates are given in Table I.

In what follows, only the 2.45 GHz band with its data rate of 250 kbit/s using O-QPSK modulation will be considered. This mode is the most popular ZigBee mode since it can be used around the world.

TABLE I. POSSIBLE COMBINATIONS OF DATA RATES AND MODULATION SCHEMES AS DEFINED IN IEEE 802.15.4.

Band (MHz) Data Rate (kbit/s) Modulation Scheme 868 20, 250 ASK, BPSK, O-QPSK

915 40, 100, 250 ASK, BPSK, O-QPSK

2450 250 O-QPSK

The different data rates result from spread-spreading methods with varying length pseudo-noise (PN) sequences. For the 250 kbit/s mode, the O-QPSK symbol rate is

c c1 2 MHzf T= = . Groups of four bits are mapped to 16 different binary chip sequences with length 32 each, yielding a spreading factor of 8SF = corresponding to a spreading gain of approximately 9 dB.

TABLE II. SPREADING SEQUENCES DEFINED IN IEEE 802.15.4

k Data Symbol

( ),0 ,3k kd d=kd � Chip Values ( ),0 ,31k k kc c=c �

0 0000 11011001110000110101001000101110

1 1000 11101101100111000011010100100010

2 0100 00101110110110011100001101010010

3 1100 00100010111011011001110000110101

4 0010 01010010001011101101100111000011

5 1010 00110101001000101110110110011100

6 0110 11000011010100100010111011011001

7 1110 10011100001101010010001011101101

8 0001 10001100100101100000011101111011

9 1001 10111000110010010110000001110111

10 0101 01111011100011001001011000000111

11 1101 01110111101110001100100101100000

12 0011 00000111011110111000110010010110

13 1011 01100000011101111011100011001001

14 0111 10010110000001110111101110001100

15 1111 11001001011000000111011110111000

The 16 different PN sequences are obtained from two base sequences which are cyclically shifted by four digits to obtain

the next sequences. The PN sequences kc for each data symbol kd are shown in Table II.

The chips ,k nc are firstly mapped to the digital baseband by applying O-QPSK modulation. Then pulse shaping with a half-sine pulse shaping filter is performed. The filter has a length of

c2T and is defined as

( ) cc

sin 0 22

0 otherwise.

t t Tp t T

π� � �≤ ≤� � �= �

��

(1)

The resulting modulation scheme is a continuous phase modulation (CPM) scheme. It can be shown that it is identical with minimum shift keying (MSK) [3].

III. MULTI DATA RATE SIGNALING The 250 kbit/s mode of IEEE 802.15.4 uses the eight times

repetition of the spreading sequence 0c as preamble. The following SFD field is fixed to ( )7 10,c c [1]. For brevity, the following shorthand notation is used for all possible combinations of two arbitrary spreading sequences.

( ), ,i j i j=s c c (2)

Since the preamble is used for synchronization, the SFD field indicates the beginning of a frame [2]. The SFD field also marks the end of the preamble and can thus also be used by the synchronizer as extra information. In order to support multiple data rates, the authors propose a scheme where modified SFD fields are used for signaling.

In order to achieve a good overall performance of the system supporting multiple data rates, the implementations used for distinguishing the different data rates should be orthogonal, what cannot be achieved with the pre-defined sequences. However, one can try to find sequences that show low pairwise cross-correlations, or large Hamming distances. Additionally, the spreading sequence 0c should be avoided to not degrade the performance of the synchronizer.

When a total of N data rates shall be supported, another 1N − SFD sequences must be found in addition to 7,10s . Let

mS denote the unordered set of N SFD sequences required to operate such a multi-rate system,

( ) ( ) ( ){ }1 2 1

7,10 p, , ,, , , , , 1Nm i j i j i jS m N

−= =s s s s� � , (3)

with pN being the overall number of permutations, i.e. different implementations of mS , when using the remaining 255 different SFD sequences except 7,10s .

( ) ( )p

255 255!1 1 ! 255 1 !

NN N N

� �= =� �− − ⋅ − +

(4)

As an example, for 7N = elements in mS , Np is in the range of 113.6 10⋅ .

978-1-4244-8132-3/10/$26.00 ©2010 IEEE

In order to obtain the optimal set of sequences a cost function has to be defined which forms the basis for the following computations. First of all, the Hamming distance h between a pair of two SFD sequences ,k ki js and ,l li js can be expressed as

( ) ( )( ) 1 1

, , , ,, ,0 0

,k l k lk l

L L

i n i n j n j ni j i jn n

h c c c c− −

= =

= ⊕ + ⊕� �s s (5)

with ⊕ representing the modulus two addition and the result being in the range [ ]0,2L , with 2 64L = corresponding to two times the length of the used spreading sequences (cf. Table II). Note that the Hamming distance in a measure for the match of a pair of two SFD sequences. The greater the Hamming distance, the smaller is the match and the lower is the chance for misdetection.

For each mS , the number of unique unordered pairs of two different SFD sequences can be expressed as

( )1

1

1 12

N

nn N NΦ

−

=

= = −� . (6)

For these Φ pairs belonging to each set mS , all corresponding Hamming distances can be collected in a vector,

( ) ( ),0 , 1 p, , , 0 1m m m mS h h m NΦ −= = −h � � , (7)

where pN such vectors exist in total.

Based on mh , various cost functions can be developed for each implementation mS of a certain unique set of N SFD sequences. Since a small Hamming distance increases the chance for misdetection, it thus refers to a high cost. Therefore, a cost vector mk associated with the corresponding distance vector mh shall be defined as

( ) ( )

( ),0 , 1

,0 , 1

, ,

2 , , 2 .m m m m

m m

k k

L h L hΦ

Φ

−

−

=

= − −

k h �

� (8)

Based on mk , two simple optimization approaches will be presented in what follows. The criterion for the first approach is the optimization of the maximum cost present in mk , and shall therefore be termed “least maximum cost” (LMC). The optimum set LMCS can be obtained by selecting

( )( )LMC min maxm

mSS = k . (9)

The criterion for the second approach is the optimization of the squared cost and shall therefore be termed “least squared cost” (LSC). The optimum set LSCS can be obtained by selecting

2TLSC min min

m mm m mS S

S = =k k k . (10)

The optimum sets LMCS and LSCS can be found by e.g. exhaustive computation following both optimization strategies for all pN permutations. They are

{ }LMC 7,10 8,8 12,12 10,7 14,5 1,1 5,14, , , , , ,S = s s s s s s s (11)

and

{ }LSC 7,10 8,3 4,14 12,9 5,6 13,1 15,11, , , , , ,S = s s s s s s s . (12)

When having a closer look at LMCS , it becomes apparent that only a subset of all available spreading sequences jc is used to form the optimum set of SFD sequences, namely 1c ,

5c , 7c , 8c , 10c , 12c and 14c . For LSCS , all sequences except 2c are used. Both optimization approaches obviously yield

completely different sets.

The optimization approach presented here was used to create a system on a chip (SoC) supporting seven data rates in the range of 31.25 kbit/s up to 2 Mbit/s, including the native ZigBee data rate of 250 kbit/s [4].

IV. PERFORMANCE RESULTS In order to prove the performance of both optimal sets

LMCS and LSCS , a simulator was established. The simulator generates frames based on IEEE 802.15.4 comprising only the preamble and the SFD field. For each element of both optimum sets, the error probability fP was determined for a signal-to-noise ratio (SNR) ranging from -10 dB up to 4 dB. An error occurs when transmitted and detected SFD field values differ. The results for the LMC approach are shown in Fig. 2 and those for the LSC approach in Fig. 3.

Figure 2. False alarm probability fP for the set LMCS .

It can be seen from the figures that the SFD field values within each optimum set perform slightly different. This property can be exploited for multi-rate systems where in general each supported data rate covers a certain SNR range. The mode operating at the lowest SNR should be signaled by

978-1-4244-8132-3/10/$26.00 ©2010 IEEE

using the SFD field with the best performance, i.e. the lowest fP , and vice versa.

Figure 3. False alarm probability fP for the set LSCS .

V. CONCLUSIONS A procedure of extending the IEEE 802.15.4 specification

for low-rate WPANs to support multiple data rates was presented. It was shown that the SFD field can act as a signaling field for the additional data rates when chosen appropriately.

In order to reuse most of the receiver components required for a native IEEE 802.15.4 receiver, the specified PN sequences have been used to form the SFD fields for additional data rates. A method was introduced allowing to select an optimum set of SFD sequences in terms of cost. In order to compute the cost for a certain set, a cost function based on the

Hamming distance between any two sequences was defined as well. Two different optimization approaches were established, both yielding an optimum set of SFD sequences.

Finally, simulation results of the error probability for each SFD field in both optimum sets of sequences were presented allowing a meaningful mapping of the SFD fields to the data rates to be supported by the extended system that shall be designed.

ACKNOWLEDGMENT The findings presented in this paper arise from a joint

project of the Department of Communication Technologies (Lehrstuhl für Kommunikationstechnik) at the University of Duisburg-Essen, Duisburg, Germany with the Ambient Technology Lab at SK Telecom, Seoul, Korea. The Department of Communication Technologies is very grateful for the strong support from SK Telecom and would like to thank SK Telecom for their continuous collaboration.

REFERENCES [1] Wireless medium access control (MAC) and physical layer (PHY)

specifications for low-rate wireless personal area networks (WPANs), IEEE Standard 802.15.4-2006, Part 15.4.

[2] S. J. Ban, H. Cho, C. Lee, and S. W. Kim, “Implementation of IEEE 802.15.4 Packet Analyzer,” World Academy of Science, Engineering and Technology, vol. 35, 2007.

[3] J. G. Proakis, Digital Communications, 4th ed., New York: McGraw-Hill, 2001.

[4] C. Spiegel, W. Shim, S. Rickers, R. Lee, G. H. Bruck, J. Yu, P. Jung, “ZigBee as a Key Technology for Green Communications,” 12th International Symposium on Wireless Personal Multimedia Communications (WPMC), September 7—10, 2009, Sendai, Japan.

978-1-4244-8132-3/10/$26.00 ©2010 IEEE