WO2006131118A1

WO2006131118A1 - Method for data communication

Info

Publication number: WO2006131118A1
Application number: PCT/DK2006/000317
Authority: WO
Inventors: Frank Fitzek
Original assignee: Aalborg Universitet
Priority date: 2005-06-07
Filing date: 2006-06-07
Publication date: 2006-12-14

Abstract

The present invention relates to a system and a method for data communication by a number of channels between transmitting and receiving means. It is the object of the invention to achieve a fast and high effective data communication. A channel number selection can be based on the data segment, which has to be transmitted, where the selected channel transmits non-data carrying signals, where the data segment can be derived in the receiver from the number of the actually selected channels. An increasing gain of nearly an order of magnitude in terms of capacity can be achieved compared to standard communication. The extremely high capacity can be very important for all kinds of communication. It is very important that also existing communication channels can be used for communication at a much higher rate.

Description

Method for data communication.

Field of the Invention

The present invention relates to a method for data communication by a number of channels between transmitting and receiving means.

Background of the Invention

Spread spectrum techniques are known from military communications. In contrast to narrow band communication, spread spectrum techniques were proved to be more resistant to jamming. For a communication system to be considered as spread spectrum system it has to satisfy the following criteria: (i) The bandwidth of the spread signal has to be greater than the information bandwidth. As this criterion is also satis- fied by frequency modulation, pulse code modulation, and delta modulation, spread spectrum has to fulfill a second condition: (ii) The spread signal is composed of the information signal and the spreading sequence. The spreading sequence has to be independent of the information [I].

In a Direct Sequence Spread Spectrum (DSSS) transmitter (or Direct Sequence Code Division Multiple Access (DS-CDMA)) the information signal is directly modulated by a spreading sequence, i.e. the generation of DSSS signals can be achieved by a simple multiplication of the information and the spreading sequence. The spreading sequence consists of a number of spreading chips with time duration T₀. The informa- tion signal consists of a number of information bits with time duration T_s. Spreading is achieved if multiple chips represent one bit. If T_s is a multiple of T_c the processing gain G can be easily calculated by G = T =T₀, Object of the Invention

It is the object of the invention to achieve a fast and high effective data communication method with high reliability for the data communication.

A further object is to achieve a high effective system for data communication using pseudo noise channel selection.

Description of the Invention

It is preferred that the preamble of claim one is modified so that a channel number selection can be based on the data segment which has to be transmitted, where the selected channel transmits non-data carrying signals, where the receiver can detect the actual number of the channel receiving the non-data carrying signals and where the data segment can be derived in the receiver from the number of the actual selected channel.

Herby can be achieved that a gain of nearly an order of magnitude in term of capacity can be achieved compared to standard speed communication transmission. The extremely high capacity can be very important for all kinds of electronic communica- tions. It should be possible for wireless communication to achieve a much higher rate of communication between all wireless communication apparatus. This could for example be mobile phones communicating over new kind of protocols. Also for wired communication this could be extremely important because the number of datasets that can be transmitted over a given wired line will increase. If this new system for com- munication is used over normal communication networks, for example over internet connections, much more data could be transmitted than is possible over existing nets. Also for optical communication over optical conducting means a new kind of data transmission will achieve a much higher rate of communication that could be very important for all kinds of telecommunication over longer distances where much more data can be transmitted over the optical line. For the whole industry it is very important if existing optical communication channels can be used for a new communication system which allows communication at a much higher rate. In a preferred for the method for data communication in a network that a number of communication channels can be established between transmission and receiving means where a spread spectrum system can be used for defining a number of communication channels, where the spread spectrum system is operating by pseudo noise spreading sequences for defining a channel descriptor for the transmitting and receiving means, where the actual data segment that has to be transmitted adjust the channel descriptor for defining the actual spreading sequence, where at least one non-data carrying bit is transmitted over the actual spreading sequence towards the receiving means, where the non-data carrying bit is received, where the data is regenerated in the receiver based on the actual spreading sequence. The use of pseudo noise speeding sequences can lead to a total use of a given bandwidth so that the whole bandwidth is used for the communication. The use of the channel descriptor for as well transmitter and receiver gives the correct coding and decoding of the signals that are transmitted. The transmitted signals are not containing any data, but simply one ore more bits de- scribing communication of a specific channel. The information is hidden in the actual number of the channels over which communication takes place. The correct coding and decoding can only take place if there is a relation between the two channel descriptors.

In a preferred embodiments for the invention the channel descriptor can be larger than the actual number of descriptors. There will be no limitation to the size of the channel descriptor, simply because the channel descriptor itself is not to be communicated over the communication net. Or if the channel descriptor is transmitted, it only has to take place now and then. Therefore very large channel descriptors can be used. Maybe channel descriptors are preprogrammed in as well transmitters as receivers and they probably do not need to be changed during life time of a product. Another possibility is, even for a large channel descriptor, that the first communication that takes place when an apparatus is switched on is exchanging or synchronization of channel descriptors. The exchange of channel descriptors over an independent media is possible, which can lead towards secret communication.

For achieving a low bit error probability spread spectrum communication system a Gaussian approximation can be used. Hereby it is possible to calculate the bit-error probability and this may be to stop communication and resend data packages if the number of bit errors is above a certain level.

The Holtzman approximation can be used for calculating the bit-error rate caused by a multiple-access interference by neglecting the effects of thermal noise for a system with equal received signal powers and randomly interfering signature sequences. Further improvements for calculating the bit-error probability can be made by using the Holtzman approximation.

Preferably the Packet Error Probability (PEP) p_p for a packet data unit of the length L [bits] can be given by Equation:

where this equation assumes a coding scheme that is used to correct bit errors in each packet. For all kinds of data communications it is very important to correct bit-errors because failures in bits that are communicated are not completely to be avoided. Many bit correction systems exist and it should be possible to use different kinds of bit correction algorithms.

Preferably a sender receiver pair has a predefined set of PN sequences for communication, where each PN sequence is referred to as a channel descriptor. The channel descriptor has to be operating or at least a correlation between the channel descriptors between transmitter and receiver has to exist. Only when this correlation between channel descriptors exists a communication without failures can take place.

In a distributed system, the PN sequences can be monitored and if some PN sequences will conflict with each other, they are replaced. In all open net work communication, for example wireless networks, there is always a risk that some parts of a communica- tion band are in use by somebody else and conflict will exist. Also in a system where pseudo noise sequences are used there exist a possibility of a conflict between transmitters. It is therefore very important that the pseudo noise sequence can be changed if necessary. This change is only possible if also information about which channel descriptor is used is communicated between transmitters or receivers.

Preferably code acquisition is used to achieve a coarse synchronization between sender and receiver. The code acquisition is very important for defining a channel descriptor. In some operating systems it would be possible that a high number of different channel descriptors exists in as well transmitter and receiver. By code acquisition the correct set of channel descriptors are used in as well transmitter and receiver. In such a system it should be possible during operation to send a new code acquisition and in that way change the whole pseudo noise sequence by changing the channel descriptor.

The receiver will know that the sender always use only one descriptor at the time, in order to decrease the impact of the synchronization, where code concatenation can be used, where the communication pair is identified by the spreading sequence C_1n. and the channel descriptor is selected by a set of codes c_t, i=l . . . H. Where the receiver tracks the ongoing transmission by de-spreading the information with c_m and post- process the channel descriptor later by sequential search using c,-, i=l, . . . , H. Hereby it can be achieved that code information is transmitted so that the correct code for the channel descriptor is always in place.

This invention also concerns a system for data communication between at least one transmitter and at least one receiver, where the communication takes place over a number of communication channels, where the transmitter comprises a first channel descriptor for selection of channels, where the channel descriptor based on an actual data segment selects a channel, which channel transmit a non-data carrying signal to the receiver, which receiver comprises a second channel descriptor, which second channel descriptor relates to the first channel descriptor, where channel number for the received non-data carrying bit is decoded in the second channel descriptor into the transmitted data segment.

Hereby a very high number of communications signals can be transmitted over a network. A spread spectrum system is used for defining a number of communication channels, which spread spectrum system is using pseudo noise spreading sequences for defining a channel descriptor for the transmitting and receiving means. Using a pseudo noise speeding sequence will make use of the whole bandwidth which is possible for the communication, and during the communication the different signals are spread over the whole area so no disturbance between the single signals occurs.

Description of the Drawing

Detailed Description of the Invention

In the following we advocate exploiting channel descriptor information in packet data communication networks to gain transmission capacity. Besides the normal data transmission also the channel descriptor (or character of the channel) can be used to convey data. This novel access technique is suitable for wired as well as for wireless networks. By the example of a wireless spread spectrum system with pseudo noise spreading sequences, we can report that a gain of nearly an order of magnitude in terms of capacity can be achieved compared to the standard spread spectrum transmission for a given scenario. Our approach is not limited to spread spectrum technologies, but applies to all systems with the property that the number of channel descriptors is larger than the actual number of simultaneously usable resources.

I. MOTIVATION

In the '60s Marshall McLuhan coined the phrase The Medium is the Message claiming that the medium over which information is transported is sometimes more important than the information itself. He was referring to the upcoming importance of television, but still 40 years later this sentence has some importance for the work we are describing in this paper. As we introduce a novel access technology, we would like to moti- vate our idea by a short example out of the GSM world. At the beginning of the GSM deployment, there was a phenomenon in Italy called squillo. People, mostly young people, would just ring each other (hanging up before the other side could pick up) using their mobile phones to say hello or to convey some other predefined messages. This kind of communication was very popular as it was not billed (still in many countries it is not). Inspired by this idea, we envision a scheme where multiple phones could be used to convey data over existing wireless networks using the signaling plane without any additional costs for the user. In Figure 1 a possible setup and example is given. The bits gained are of course paid for by the network provider and far fewer bits are conveyed than the network provider has to invest to make this transmission possible. Therefore it may be referred to as a trick. Leaving this example, we raise the questions whether it is possible to convey data by exploiting the channel descriptor (in our example phone numbers) used instead of transmitting data (recall that we did not send any bits at all over the GSM bearer). In the following we would like to investigate this idea in more detail. We will find out that the proposed approach will not lead to any gain if the number of channel descriptors is less or equal than the actual resources that can be used simultaneously. Therefore we are not looking into orthogonal systems but at so called spread spectrum systems with pseudo noise sequences, where the number of descriptors (all available spreading sequences) is larger than the descriptor itself.

Fig. 1. Data Transmission over the GSM signaling plane with two groups of four mo- bile phones. Both groups have one master (M) each and three slaves (S) communicating over Bluetooth among group members. One sending phone, acting as a master, has the information to be conveyed to the master of the receiving group. The four sending and receiving phones are identified by their phone numbers and to each entity a two bit address is assigned. This assignment is known to both masters of the group. The sender master will read the digital message and by using the first two bits one phone of the sender group is chosen to call over GSM a phone of the receiver group, which is identified by the second two bit tuple. In this example, fist the master itself (00) will call the second phone of the receiving group (01). The receiving phone, by using the intra group communication, informs the master about the received call (also which phone in the sending group made the call), which the master in rum can demap into four bits of information (0100). The second call from phone number four of the sender group (11) to the third phone of the receiver group (10) will be transformed into the information (1011). By each call four bits of information are transmitted. The remainder of the text is structured as follows: In Section II a sophisticated survey on spread spectrum technology extracting the information to explain our novel approach is given. The novel approach itself is explained in Section III; first for the ideal receiver, then for the non-ideal receiver case. A first performance evaluation of our approach is given in Section IV. We discuss our approach in Section V and conclude the work by Section VI.

II. SPREAD SPECTRUM TRANSMISSION

Spread spectrum techniques gained their popularity by the needs of military communications. In contrast to narrow band communication, spread spectrum techniques were proved to be more resistant to jamming. For a communication system to be considered a spread spectrum system it has to satisfy the following criteria: (i) The bandwidth of the spread signal has to be greater than the information bandwidth. As this criterion is also satisfied by frequency modulation, pulse code modulation, and delta modulation, spread spectrum has to fulfill a second condition: (ii) The spread signal is composed of the information signal and the spreading sequence. The spreading sequence has to be independent of the information [I].

In a Direct Sequence Spread Spectrum (DSSS) transmitter (or Direct Sequence Code Division Multiple Access (DS-CDMA)) the information signal is directly modulated by a spreading sequence, i.e. the generation of DSSS signals can be achieved by a simple multiplication of the information and the spreading sequence. The spreading sequence consists of a number of spreading chips with time duration T_c. The informa- tion signal consists of a number of information bits with time duration T_s. Spreading is achieved if multiple chips represent one bit. If T_s is a multiple of T_c the processing gain G can be easily calculated by G = TJT_C.

Fig. 2. Standard Spread Spectrum Transmission Approach.

The proper choice of spreading sequences enables multiple access capability for spread spectrum based wireless communication systems. A sequence is a non ambiguous identification for a transmitter receiver pair. Spreading sequences can be divided into orthogonal and Pseudo Noise (PN) spreading sequences. In this paper we focus on PN sequences. PN sequences are binary sequences, which exhibit random properties similar to noise. Within the class of PN sequences, the most popular representatives are Maximal Length sequences, Gold sequences, and Kasami [2] sequences. All sequences can be generated using a Linear Feedback Shift Register (LFSR), which is built by /feedback taps [3]. Sequences generated with a LFSR having the maximum possible period length for an/- stage shift register are called Maximal Length or simply m-sequences. The length of an m-sequence can be proven to be 2/- 1. The number of possible codes depends on the number of possible sets (also called primitive irreducible generators) of feedback taps. Golomb [3] showed that the overall number of sequences generated by a LFSR of degree/equals

where PJ equals the prime decomposition of 2^f~ 1.

Corresponding to N_1n for the m-sequences, we give the number of spreading sequences for the Gold and Kasami sequences in Equation 2 and 3, respectively.

N₉(f) = 2> H- 1, (2)

^{m (f) ~}

⁽³⁾

Table I provides the sequence length of a given degree /and the resulting number of achievable code sequences (see [4] for higher values off). TABLE I

SPREADING SEQUENCE LENGTH WITH RELATED NUMBER OF MAXIMAL

LFSR (DEGREE/) SPREADING SEQUENCES.

The Bit Error Probability (BEP) p_s for a standard multi-user spread spectrum communication system are given in Equation 4 using a Gaussian approximation. The Gaussian approximation does not reflect the single auto- and cross correlation among the used spreading sequences. But in real systems, using PN sequences, the communica- tion pair agrees on a very long PN sequence. Shorter PN sequences are extracted from this long PN sequence periodically and used for the spreading. By this approach persistent interference between terminals, which would occur if two terminals were using the same series of PN sequences, is avoided and therefore the Gaussian approximation can be used.

It can be seen that the BEP p_s depends on the system parameter G (spreading factor) and the number of users k communicating at the same time. An improved Gaussian approximation has been derived by Holtzman [5] and applied in [6]. The Holtzman approximation calculates the bit-error rate caused by the multiple-access interference (neglecting the effects of thermal noise) for a system with equal received signal powers and randomly interfering signature sequences. The bit error probability for the improved Gaussian approximation is given in Equation 5. The calculations for the bit error probability are still simple enough but lead to quite accurate results. In [7] it is claimed that the improved Gaussian approximation should be used, if the number of active connections is small or the spreading gain G is large.

PHoltzman(k) ~ Q Q

α³ = (fc - I) (^G^S + (X ₊ J^) (G - I)) . (5)

The Packet Error Probability (PEP) p_p for a packet data unit of the length L [bits] is given in Equation 6. This equation assumes a coding scheme that allows us to correct e bit errors in each packet [8], [9].

Equation 7 [10] gives the capacity C for a single CDMA cell as a function of the packet length L, the redundancy R (to correct e bits) and the number of active wireless terminals k.

C = (L - R) - U - (I - P_p) (7)

The cell capacity is depicted in Figure 3. Based on the bit error probability, we calculate the packet error probability by considering a simple static FEC as follows. We set the packet length to 1023 bits and employ static forward error correction that can correct up to 30 bit errors. We assume a slotted channel structure. The spreading gain is set to 32. For a small number of active channels the capacity increases linearly. At a specific point the throughput decreases rapidly. The improved Gaussian approximation by Holtzman has an optimal number of seven active channels for this example.

Fig. 3. p, P_p and Capacity versus the number of active users. The optimal number of active users (for G=32, L- 1023, R=295, e=30) is 7.

Thus, the admission control should take care that not more than seven channels are used. But at the same time, using Gold sequences, a total number of 33 spreading se- quences are available. The observation that the number of channel descriptors is larger than the number of resources that can be used simultaneously is exploited in our proposed scheme. Note, that for orthogonal systems the number of channels descriptors is always the same as the number of resources used simultaneously.

III. NOVEL APPROACH FOR SPREAD.SPECTRUM TRANSMISSION

In our proposed scheme we assume that a sender receiver pair has a predefined set of PN sequences for communication. Each PN sequence is referred to as a channel descriptor. This approach is similar to a Multi Code CDMA (MC-CDMA) system. In contrast to a MC-CDMA system our approach will always use only one channel at the time. In Figure 4 the novel spread spectrum transmission approach is given. If the number of parallel channels per communication pair is denoted as H and the binary message s has to be conveyed to the receiver, the sender will take the first N_d = 1Og₂H bits of the binary sequence s identifying the channel descriptor to use and send Ns bits over this channel descriptor. Note that the N_d are not sent over the air interface and therefore are not introducing interference to other ongoing transmissions. On the receiver side the ongoing communication on one of the channel descriptors is detected and the N_s bits are received. Before the N₁ bits are written into the receiver queue the receiver will add N_d bits, known by the channel descriptor used. From the perspective of ongoing transmissions in the cell, the proposed system will produce interference like a single code CDMA system. As CDMA is interference limited, we claim that our proposed scheme will have no effect on ongoing transmissions and can therefore be evaluated in a stand alone fashion. If we consider multi user scenarios and all receivers are merged in a centralized entity such as a base station then double use of PN se- quences among the terminals can be avoided easily. For a distributed system, the PN sequences have to be monitored and if some PN sequences will conflict with each other, they should be replaced.

Fig. 4. Novel Spread spectrum Transmission Approach.

For illustration purpose, we assume H = 8 such that b = 1Og₂N = 3 bits are assigned such as given in Table IL Furthermore only N_s = 2 bits are transmitted over each channel descriptor at the time and for this time we assume that the spreading code acquisition is perfect. In case the sender has to transmit the binary sequences s - (001111010100100) it would take spreading sequence C₀₀₁ to transmit the bit value 11 in the first slot (using this spreading sequence identifies the first three bits and the value transmitted the bit number four and five), the spreading sequence C₁₀₁ to transmit the bit value 01 in the next slot and the spreading sequence C₀₀₁ to transmit the bit value 00 in the last slot.

TABLE II

SPREADING SEQUENCES ASSIGNMENT FOR H=8.

spreading sequence bits assigned

1 cooo 000

2 COO l 001

3 coio 010

4 con 011

5 cioo 100

6 ClOl 101

7 ClIO 110

8 cm 111

' In this example we have only transmitted six bits over the wireless channel, but the receiver receives 15 bits of information. Thus the capacity was improved by the factor of 2.5 in contrast to a system where only one channel would have been available transmitting all 15 bits. In general the gain Φ is defined as the ratio of bits sent with the new scheme divided by the bits sent in the old scheme as given in Equation 8. bits _old

A. Ideal Spread Spectrum System

In case the spreading code acquisition and synchronization is ideal, the gain can be simply calculated as.^{φ = "}"⁰NT^- \ in the ideal case the capacity of each terminal for the proposed system is Φ times higher than the standard one. The bit error rate remains the same as the cell interference will not change in contrast to the standard transmission case for the transmitted bits. The approach can be also used to reduce the inference by transmitting as much information as before, but transmitting less information over the air interface exploiting our approach. In such a case we can assume that the number of active users is simply divided by the gain Φ . Furthermore, in the ideal case, only the JV₅ bits are prone to errors, while the N_d bits are error free. Therefore the bit error probability has to be divided by Φ .

For the ideal case the BEP versus number of parallel channels and versus active users for the standard and the proposed approach is given in Figure 5. For the proposed scheme we have depicted the bit error probability for the transmitted bits and the overall bits individually. We used the simple Gaussian approximation here. It can be seen that our proposed scheme yields significantly better results than the standard approach. Without going into details we leave the ideal case and investigate the more realistic scenario of the non-ideal receiving process. B. Non Ideal Spread Spectrum System

So far we assumed the spreading code acquisition and synchronization are perfect leading to an error-free reception of all N_d bits. In real systems we have to deal with code acquisition and synchronization. Here and hereafter we use the

Fig. 5. BEP Versus Number Of Parallel Channels And Versus Active Users For The Standard And The Proposed Approach For The Ideal Case Using The Gaussian Approximation.

Holtzman approximation to calculate the bit error probability. Code acquisition is needed to achieve a coarse synchronization between sender and receiver. In case the code acquisition fails, no data will be transmitted at all. Two different types of acquisition errors are known, namely a false alarm (incorrect hypothesis to have received a signal) and a miss (not successful detection of a send signal). Possible detection schemes are the matched filter energy detector and the radiometer. Both approaches are comparing the incoming signal σ with a given threshold γ. Whenever σ is larger than γ a detection is assumed. A large γ will end up in less detection but also less failure and the other way round. The probability for a false alarm p_fa and a miss/?_OT (or its complement the probability of detection p_de) are given in [11], [12], [13] for matched filter energy detector and the radiometer assuming a single code CDMA system. We give here the probabilities for the radiometer as:

To calculate the non-ideal BEP/?,,. of our approach with code acquisition and synchronization errors the fraction of erroneous bits by the total number of bit has to be calculated as:

where p_} = p_r ^■ (1 - p_f)^HΛ reflects the situation where a signal is detected only on one channel with probability/?,, (probability that the signal is detected on the correct channel after the system has been synchronized) and the other channels have no false alarm meanwhile with probability p_f (probability that a false signal is detected on one chan- nel after the system is synchronized). The probabilities set p_r and p_f are much smaller than the set of probabilities p_de and ρ_fa, respectively, as the later set gives the probabilities when no synchronization has been established yet. The former set includes the initial synchronization and the ongoing tracking. For completeness, the probability 1 - p_λ reflects the situation, where multiple detections are assumed, no de- tection has occurred, or the transmission is not synchronized anymore. For the standard approach the error prone detection is also taken into consideration as given below

where p_x =p_r (so no false alarm will worsen the situation).

To decrease the impact of the code acquisition error, we advocate a joint detection (JD) of the signal. So far we left it up to each channel to compare the incoming signal strength σ with a given threshold γ individually to check for detection, a situation which we refer to as single detection (SD). In contrast to the SD approach we take the highest value of σ of all channels and take this as a detection, referring to this as the JD approach. This is possible as the receiver knows that the sender will always use only one descriptor at the time. To decrease the impact of the synchronization, code concatenation can be used as proposed in Figure 6. In this case the communication pair is identified by the spreading sequence c_wτ and the channel descriptor is chosen by a set of codes C₁, i=l . . . H. The receiver can then track the ongoing transmission by de-spreading the information with c_m and postprocess the channel descriptor later by sequential search using C₁, i=l, . . . , H. We claim that this setup is less error prone as the de-spreading with C_1n will take place H times more often than on separate multiple channels. Therefore the synchronization can be maintained more easily as the clock of c_m can be used by the later de-spreading with c_t. Furthermore when we detect a signal with C_fn, we can be sure that one and only one of C₁, will detect the signal in the second level. With this approach the error probability of the transmitted bits is equal to that of the standard one.

Fig. 6. Novel Spread.spectrum Transmission Approach With Code Concatenation.

With the new architecture the probability p_f is set to zero, Equation 14 has to be re- written to

(1 - Pr) (N_d + N_s(l - p_s))_Pl p_n» = 1 - (14) 2 N_s + N_d

IV. PERFORMANCE EVALUATION

For the performance evaluation we assume JV = 99.9% and p_f — 0.1% with a spreading gain G = 32. Up to e = 30 bits can be corrected in a L = 1023 bit word using Fig. 7. BEP Versus Number Of Active Users For The Standard And The Proposed Approaches With Code Acquisition Errors.

Fig. 8. Capacity Versus Number Of Active Users For The Standard And The Proposed Approaches. For The Standard And The Proposed Approaches.

e = 30 bits can be corrected in a L = 1023 bit word using R = 295 bit of redundancy. We analyze the different number of H= 1; 2; 4; 8 leading to values of N_d = 0; 1; 2; 3. N₁ is set to one. The standard approaches (ideal and non-ideal) are given by solid lines, while our proposed approaches are reflected by dashed and dotted lines. The single detection case (SD) has non-filled points, while the joint detection (JD) is given with filled points. In Figure 7 the BEP versus the number of active users for the standard and the proposed approaches with code acquisition errors are given. In that compari- son also the standard communication system is prone to errors due to the code acquisition. In Figure 7 the BEP curves of the JD approach are the best and the more channels that are available, the better the performance will be. It seems that the performance is only limited by the number of available descriptors per communication pair. The non ideal standard approach yields better results in terms of the BEP values than the SD approach as the number of users are smaller than 6, 9, and 12 for H equal 2, 4, and 8 respectively. In Figure 8 the capacity versus number of active users for the standard and the proposed approaches are given. Here a more detailed discussion about the performance can be made. Obviously the JD approach yields better results than any other approach. The performance increases with an increasing number of H as for the BEP. The SD approach yields the best performance for H = 2, as a larger number of H will degrade the performance due to many false alarms. For H = 8 the performance is even worse than the standard approach (for fewer than 11 users). The maximum cell capacity is achieved with six users in the cell for the standard and with seven users for the SD approach. For the JD approach the optimal number of users is 8, 10, and 11 when H equals 2, 4, and 8 respectively.

For illustration purposes we present two kinds of cell gain. Gain_s defines the cell gain of a given approach divided by the standard approach with the same number of users, while Gain_sopt defines the cell gain of the proposed approach divided by the capacity of the standard approach with each approach using its own optimal number of users. The results are given in Table III. As we can see, Single Detection reaches its maximum of 2.16 when H= 2, and Joint Detection reaches its maximum of 7.46 when H- 8.

TABLE πi

GAIN AT THE OPTIMAL WORKING POINT.

no. users Approach H Capacity [bits] Gains Gain_{s >opt}

6 Standard 1 3497 1.00 LOO

7 SD 2 7425 2.36 2.16

7 SD 4 5775 1.80 1.65

7 SD 8 724 0.22 0.21

8 JD 2 9960 3.99 2.84

10 JD 4 17437 20.34 4.98

11 JD 8 26101 71.34 7.46

V. DISCUSSION

In the discussion two points are highlighted, namely whether our system still meets all of the requirements for a spread spectrum system and the possibility of changing the PN sequences on the fly. Referring to the definition of the spread spectrum system given in Section II, the spreading information and the transmitted information should be uncorrelated. This requirement still holds as the transmitted information has no impact on the choice of the spreading code. The second point deals with the feasibility of changing the PN sequence on the fly. Here we are referring to the implementation in an IS-95 [14] system, where long PN sequences are assigned and subsets of this long PN sequence are extracted to spread the information streams. While in IS-95 the subsets are taken out step by step, we are modulating our information onto the choice of smaller PN sequences. Therefore we claim that our approach is as complex as any standard system and that changing PN sequences is a solved problem. Furthermore in IS-95 systems the usage of multiple channels was already addressed by the LIDA [15] and BALI [ 16] approach. VI. CONCLUSION

We have introduced a new access technique based on the exploitation of channel descriptor information. We are exploiting the medium over which partial information is conveyed to code additional information. Following McLuhan, we say the medium is the message or the medium is part of the message. By the example of spread spectrum we have shown the way our approach works and have given a first performance evaluation. The cell gain will vary depending on the system but in our example, using realistic receiver architectures, our technique gave an improvement of nearly one magnitude over the standard one (both with the optimal number of users). In this paper we used the spread spectrum technology to illustrate our approach. Other technologies can be found and will be named here in our final paper. In our future work we will apply this approach to further communication systems which are not limited to spread spectrum technology.

Claims

1. Method for data communication by a number of channels between transmitting and receiving means, characterized in that a channel number selection is based on the data segment which has to be transmitted, where the selected channel transmits non- data carrying signals, where the receiver detects the actual number of the channel receiving the non-data carrying signals and where the data segment is derived in the receiver from the number of the actual selected channel.

2. Method according to claim 1 characterized in that the method for data communication in a network where a number of communication channels are established between transmission and receiving means, where a spread spectrum system is used for defining a number of communication channels, where the spread spectrum system is operating by pseudo noise spreading sequences for defining a channel descriptor for the transmitting and receiving means, where the actual data segment that has to be transmitted adjust the channel descriptor for defining the actual spreading sequence, where at least one non-data carrying bit are transmitted over the actual spreading sequence towards the receiving means, where the non-data carrying bit is received, where the data is regenerated based on the actual spreading sequence.

3. Method according to claim 1 or 2, characterized in that the channel descriptor is larger than the actual number of channel descriptors.

4. Method according to one of the claims 1-3, characterized in that for achieving a low bit error probability by using spread spectrum communication system a Gaussian approximation is used.

5. Method according to claim 4, characterized in that the Holtzman approximation is used for calculating the bit-error rate caused by a multiple-access interference by ne- glecting the effects of thermal noise for a system with equal received signal powers and randomly interfering signature sequences.

6. Method according to one of the claims 1-5, characterized in that the Packet Error Probability (PEP) p_p for a packet data unit of the length L [bits] is given by Equation:

where this equation assumes a coding scheme that is used to correct bit errors in each packet.

7. Method according to one of the claims 1-6, characterized in that a sender receiver pair has a predefined set of PN sequences for communication, where each PN sequence is referred to as a channel descriptor.

8. Method according to one of the claims 1-7, characterized in that in a distributed system, the PN sequences are monitored and if some PN sequences will conflict with each other, they are replaced.

9. Method according to one of the claims 1-8, characterized in that code acquisition is used to achieve a coarse synchronization between sender and receiver.

10. Method according to one of the claims 1-9, characterized in that the receiver knows that the transmitter will always use only one descriptor at the time, for decrease the impact of the synchronization, code concatenation is used where the communication pair is identified by the spreading sequence c_m and the channel descriptor is se- lected by a set of codes c_t, i=l . . . H. where the receiver track the ongoing transmission by de-spreading the information with c_m and postprocess the channel descriptor later by sequential search using C₁, i=l, . . . , H.

11. System for data communication between at least one transmitter and at least one receiver, where the communication take place over a number of communication channels, characterized in that the transmitter comprises a first channel descriptor for selection of channels, where the channel descriptor based on an actual data segment selects a channel, which channel transmit a non-data carrying signal to the receiver, which receiver comprises a second channel descriptor, which second channel descriptor relates to the first channel descriptor, where channel number for the received non- data carrying bit is decoded in the second channel descriptor into the transmitted data segment.

12. System according to claim 11 characterized in that a spread spectrum system is used for defining a number of communication channels, which spread spectrum system is using pseudo noise spreading sequences for defining a channel descriptor for the transmitting and receiving means.