By Dick Caro, CMC Associates
Karlsruhe, Germany – The SP100.11a standards working group met here for four full business days to make progress toward the standard. By the end of the meeting, the committee had agreed upon an architecture, selected a single application layer and formed the editorial team to actually write the standard.
While cooperation among competing suppliers seems to be working very well, there continues to be a feeling that each of the two groups—CI (Collaboration Initiative, lead by Emerson) and WNSIA (Wireless Network for Secure Industrial Applications, lead by Honeywell)—will try to gain an advantage. After last month’s user objections, the committee agreed that the Honeywell attempt to get its narrowband frequency-hopping spread spectrum physical layer adopted as an alternative would be deferred to Release 2. Meanwhile, Release 1, based on the CI recommendation using an IEEE 802.15.4-2006 radio with channel-hopping, continues to take shape.
The architecture of the SP100 network has been the subject of many heated discussions. The architecture adopted for Release 1 is a full mesh network in which all nodes may contain a routing function, but it can be disabled or omitted for simplicity or power conservation reasons. The mesh allows for a network to become physically large without installation of a higher-speed backbone, but each “hop” comes with a time delay that may prevent the use of the data from far points in fast control loops. This is why Release 1 is targeted at non-critical functions that can tolerate a latency (time lag) of 100 milliseconds. This limitation is often expressed as a control loop cycle time of about one second.
The Honeywell alternative architecture proposed at the Phoenix meeting included a wireless backbone and field-mounted SP100 access points that were interconnected using the IEEE 802.11s (still in draft stage) wireless mesh protocol. This architecture will most likely emerge in Release 2 of SP100.11a as an option. Honeywell has shown that its proposed architecture is more time-efficient than the full mesh network proposed for Release 1, and is more suited for faster control loops and even safety systems. However, it was agreed that nothing in the architecture for SP100, even in Release 1, would prevent operation with latencies as low as 10 milliseconds.
After presenting a proposal for the application layer (AL) in Phoenix, Honeywell delivered an almost complete document for consideration to the Application/Gateway Task Group in Karlsruhe. This AL is highly compatible with Foundation fieldbus, HART and Profibus, since it is based on the use of electronic device descriptions (EDD) taken from IEC 61804-3, an international standard now being balloted as ISA/ANSI 104.
SP100 AL will not specify any EDDs, but will allow the use of EDDs by applications connected through an SP100 network. All of the required AL services are included in this document, including a tunneling or encapsulation protocol for transmitting messages other than SP100. The remaining committee work is to convert the Honeywell document to the correct form for a standard. This committee also heard an alternative AL presentation from Rockwell Automation for its CIP protocol. It was generally agreed that CIP would be considered for a later release that was more oriented toward factory automation; however, CIP will be reviewed to see if any of its features are suitable for inclusion in Release 1.
The committee is planning to include a gateway function for SP100.11a. The gateway may be contained in a separate device, or it may be distributed to other network devices. The gateway is designed to translate from SP100 data streams to those of the supported networks—Foundation fieldbus, Profibus/PROFInet or HART. SP100 will not specify the translation utility, but will provide the network interfaces required so that the respective organizations can create one.
Since HART is using a wireless network totally incompatible with SP100, we cannot plan that that these signals can be communicated via the SP100.11a network in any way. All that we can do for wireless HART is to specify the gateway functions that a device could use to be compatible with the SP100 interface to the host control system.
The work of the Network/Transport Task Group has now been defined. In SP100, the mesh routing (intranet) will not be done in the network layer (NL), but is assigned to the upper data link layer. The NL will only handle routing between different SP100 network segments. The services that the transport layer (TL) will provide to applications are cyclic and acyclic reads and writes, alerting and multicast messages. The requirements for these services were obtained from the draft AL specification.
The exact protocol for the NL has not yet been formalized. Since the NL must allow routing between connected SP100 networks, it must provide routing services in the protocol stack. One such stack proposal is called 6LoPAN, which is designed to use a compressed version of Internet Protocol, version 6, over a low-powered, low-data-rate network. 6LoPAN has recently been approved by the Internet Engineering Task Force (IETF), the Internet standards body, as a draft standard for low-powered networks. A decision on the NL protocol must be made soon.
Release 1 will have only one physical layer (PHY). This is similar to the original proposal by CI group members Dust Networks and Emerson. It will operate at 2.4 GHz using direct-sequence spread spectrum with added channel-hopping. This PHY will use IEEE 802.15.4-2006 radios.