We address an issue of channel sharing among users by using a random assignment/transmitter-oriented (RA/T) code scheme which permits the contention mode only in the transmission of a header while avoiding collision during the data packet transmission. Once the header being successfully received, the data packet is ready for reception by switching to one of programmable matched-filters. But the reception may be blocked due to limited number of matched-filters so that this effect is taken into account in our analysis. We also consider an acknowledgement scheme to notify whether the header is correctly detected and the data packet can be processed continuously, which aims at reducing the interference caused by unwanted data transmission. For realistic analysis, we integrate detection performance at the physical level with channel activity at the link level through a Markov chain model. It is shown that compared to classical CDMA system, about half reduction in receiver complexity is allowed by choosing a proper number of RA/T codes without losing performance quality in view of the normalized throughput.
Most of direct-sequence/spread-spectrum multiple-access (DS/SSMA) packet radio networks are usually allowed to change their spreading code sequences while transmitting. If so, system throughput and complexity are largely affected by selection of spreading code sequences. In spread-spectrum networks, transmission protocol as a rule of determining such selection is specified by a number of factors, such as transmitter, intended receiver, transmission time, and priority of message, etc. Up to now, there are several studies [1], [2] to address this issue, but a few results [3], [4] have been reported to provide detailed analysis of performance.
This paper proposes a random assignment/transmitter-oriented code scheme for centralized DS/SSMA packet radio networks. Here RA/T refers to selection of two spreading codes to be used for transmission of the header and data portion of a packet. When a terminal is ready to send a packet to a central node, it chooses randomly one out of spreading codes [4] for transmission of the header in which is considerably smaller than the number of users. The data packet is then transmitted using a distinct spreading code to avoid collision among contending packets. In the RA/T code scheme, correct detection of the header mainly determines system throughput, which enables us to continuously process the data packet by switching to one of programmable matched-filters. But if we consider much less than to reduce system complexity, some of data packets may be blocked even though their preceding headers are correctly detected. System throughput is then affected by a complicated function of detection performance at the physical level and channel activity at the link level.
In the RA/T code scheme, a number of terminals may transmit their headers using the same spreading code which will cause collision leading to failure reception of these headers. Here the multi-user interference among them is referred to as the primary user interference. If the time delay between the first two arrivals is sufficient to distinguish between them, the first one may be captured [5], [6] using a high time resolution property of direct-sequence spread-spectrum signals after taking correlation. On the other hand, the multi-user interference caused by simultaneous transmissions using distinct spreading codes is referred to as the secondary user interference, which does not involve collision because of the multiple- access capability of our system and increases only the bit error rate.
For evaluation of system throughput, we obtain an approximation to the probability of minipacket success by taking into account the bit-to-bit error dependence within a minipacket [7], [8], and model the network state as a Markov chain. At the link level, we account for a collision event caused by the primary user interference and a blocking event due to the limited number of matched- filters in deriving the state transition probability of the Markov chain. At the physical level, the effects of the secondary user interference and error-correction coding are considered when we evaluate the probability of minipacket success. Thus, system throughput normalized by a code rate and its bandwidth shows the performance of the RA/T code scheme which reflects the characteristics of physical level and also the complexity of central node.
For the RA/T code scheme, a central node can not receive continuously the data packet unless its preceding header is correctly detected, since the central node identifies a source address used for data demodulation by decoding the header. Also, in case of , the number of successful transmissions in a slot can not exceed so that the data packet may not be received whenever all matched-filters are preoccupied for data reception. If a terminal is transmitting such unsuccessful data packet, then it will increase only the secondary user interference. In order to remove unwanted interference, we may adopt an acknowledgement scheme to inform a receiving status immediately after processing the header. In this case, the terminal waits an acknowledgement after sending the header of a packet and then send the following data packet if the acknowledgement is received. Otherwise, it will stop sending and attempt retransmission later, which allows to reduce the interference and increase the multiple-access capacity of the proposed RA/T code scheme.
The organization of the paper is as follows. In Section II, we provide an overview of channel sharing using the RA/T code scheme along with system model. An accurate, simple Gaussian method is applied to obtain the approximation for the probability of minipacket success in Section III. In Section IV, an overall throughput is analyzed by accounting for detection performance and channel activity, and the analysis is extended to the case of using an acknowledgement in Section V. Numerical results are presented in Section VI with concluding remarks.