With the increasing complexity of cabling at sensory level in process control and manufacturing automation, field buses were introduced to replace the traditional point to point links from each sensor or actuator to its controlling equipments by a single link on which all information is transmitted serially and multiplexed in time.
In this thesis work, we introduce the BITBUS network as a field bus and its performance is studied. For the service discipline of the BITBUS network protocol, two service strategies are defined to obtain the performance of the BITBUS network. They are the equal priority cyclic service strategy and the non-equal priority cyclic service strategy. The former assigns equal priority to each node for polling and the latter assumes non-equal priority.
For mathematical analysis of the BITBUS network performance, the BITBUS network was modeled as a cyclic queueing model. In that model, the poisson arrival and the equal priority cyclic service strategy are assumed. The cyclic queueing model was analyzed by two methods: the Kuehn's method and the Boxma's method. Computer simulation was also done for the cyclic queueing model and the simulation results were compared with those of Kuehn's and Boxma's. Close agreements between the simulation results and the analyses were obtained at the light traffic condition of practical interest. Under the heavy traffic condition, the Boxma's analysis method yielded superior results to the Kuehn's method when each node has the same traffic intensity.
In order to obtain the performance of the BITBUS network under the condition of the non-equal priority service strategy or the periodic message arrival, only the computer simulation was done. The simulation results show that the waiting time was maintained stable up to the critical traffic point if the periodic message arrival was assumed. When each node has the different traffic intensity, the non-equal priority cyclic service strategy was superior to the equal priority cyclic strategy. In reality, in order to meet the response time requirement of 5 msec imposed by International Electrotechnical Commission when each node has the average traffic of 5000 messages/sec in manufacturing automation, the number of slave nodes should be smaller than 10 at the transmission rate of 2.4 Mbps.
Finally, a brief description of the BITBUS hardware and an interface method between the BITBUS network and Mini-MAP are presented.
공정 제어나 제조 자동화에 있어서 현장의 각 sensor나 actuator와 controller 간의 연결에 있어서 종래에는 point to point 연결로 인해 cabling 문제가 점점 복잡해졌는 데 이 문제를 극복하기 위해 한 개의 연결로 대체하고 각 sensor나 actuator가 time multiplexing에 의해 한 선을 공유하는 Field Bus를 도입한다.
본 논문에서는 Field Bus로 구현된 BITBUS network을 소개하고 성능을 분석한다. BITBUS protocol에서 정의되지 않은 cyclic service strategy로써 모든 node에 같은 priority를 주는 균등 priority 방식이나 특정한 node에 priority를 주는 차등 priority 방식을 고려한다.
균등 priority 방식과 poisson arrival process를 도입한 상황에서 Kuehn의 방식과 Boxma의 방식을 이용하여 수학적 해석을 하고, 같은 상황에서 simulation을 행하여 서로의 결과를 비교해 본다.
결과에 의하면 high traffic에선 Kuehn 방식과 Boxma 방식 모두 simulation 결과와 차이를 보이고 있고 low traffic에선 비교적 일치함을 보여주고 있다. Traffic이 비대칭인 상황에선 Boxma 방식과 Kuehn 방식이 비슷하지만 대칭일수록 Boxma 방식이 Kuehn 방식보다 더 simulation에 접근함을 보여주고 있다.
다음으로, 수학적 해석은 안 되지만 실제 환경에 보다 가까운 환경으로 주기적 message 도입과 차등 priority 방식을 추가로 도입하여 simulation 하였다.
결과에 의하면 주기적인 message 경우 임계 traffic 이전까지는 안정된 waiting 시간을 보여주고 있고, traffic이 특정 node에 편중될 경우 차등 priority 방식이 유리함을 보여주고 있고, IEC에서 요구하는 제조자동화에 있어서 응답시간 5 msec내에 이르기 위해선 2.4 Mbps의 전송률로 10개의 node까지 연결할 수 있음을 보여주고 있다.
마지막으로, BITBUS hardware를 간단히 묘사하고 BITBUS Network과 Mini-MAP과의 Interface 문제를 개념적인 수준에서 묘사하고 있다.
