There are 26 #Table# markers remaining in the specification document.
These need to be replaced with their proper numbers and subsequent tables renumbered.

Resolve Table number placeholders and renumber

Each of the 26 #Table# placeholders for table numbers in the specification document should be resolved to its proper sequential number (or omitted in the cases where there is no table) and subsequent tables renumbered accordingly.
Also remove heading for non-existent tables (e.g. for typedefs which have no attributes).

The GraphQL mapping in the specification and defined in the schema files leads to an unnecessarily complicated implementation of the standard and should be simplified

Regenerate GraphQL PSM with updated mapping

The updated mappings from PIM to GraphQL as incorporated into the TDAI beta spec and C2INav v1.1 RTF should be used. This generates a single schema for each standard avoiding class duplication, has a flatter structure with respect to endpoint methods, has less duplication with respect to input types and uses a simpler mapping with less duplication for inheritance (consistent with that also now used for DDS/IDL - see OARIS2-27). The proposed schema file is attached.

Heartbeat_Signal.idl has been removed from the distribution. Searching for receive_subsystem_heartbeat_signal and receive_cms_heartbeat_signal in the IDL returns no entries. The interface can’t be found.

Exchange Heartbeat is documented as a Compliance Level 2 interface in the IFS. It is documented at section 7.7.5.2. It can only be concluded therefore that the Heartbeat_Signal.idl file has been omitted by mistake.

It is noted that the documentation omits subsystem_id as per the other interfaces. This is an error?

Document omission of Heartbeat Signal as in-built in DDS PSM

In section 8.2 describe how Heartbeat Signal is also mapped to DDS in-built API methods (as well as Register Interest, already documented).

plot_count field of concentration_plot_cell_type is defined as unsigned long long 64-bit type, which causes issues with implementation languages and machine base type representation on 64-bit machines.
Is it viable for a count of the number of plots in this clutter reporting context to exceed 2**32 (4 billion)? Would unsigned long be sufficient, hence making implementation more machine-friendly?

Change type of plot_count attribute

The plot_count attribute in struct concentration_plot_cell_type should be an unsigned long (rather than unsigned long long). This has more than sufficient range of values for any application of the clutter reporting use case.

The Plot Reporting Use Case is confusing. For OARIS 1.1, the actual Plot structure was abstracted into a sub-type write_sensor_plot_type defined in Provide_Plots.idl. In OARIS 2.0, the IDL file Provide_Plots.idl has been removed, with the interface effectively reverting back to the OARIS 1.0 version with sensor_plot_type defined in Plot_Reporting.idl. Is this intentional or was an incorrect baseline used for creating OARIS 2.0? This may also explain why subsystem_id is also missing as per sensor_track_type.

Flattened Structure Is Intentional

This change is a consequence of applying the principal of flattening PIM methods with a single parameter consistently throughout the whole model when mapping to the DDS/IDL PSM (and also GraphQL).

The topic type for Plots should have been flattened in the same way as that for Tracks in v1.1. This has now been done for v2.0.

The mapping of specialisations duplicates the location of common attributes from the generalisation class into each option of the union. This is inconvenient for the software design of applications using the types.

Use a rationalised DDS/IDL inheritance mapping.

Update the mapping of generalization / specialization inheritance so that attributes of the generalization class are encapsulated into a base class, which then appears as an attributes of each specialization rather than each of the generalization attributes individually.
Also the generated enumeration values are missing the name of the generalization class in their IDL documentation comment.

The impacted IDL files are attached as per the proposed update.

The following attributes are not correctly linked in the XMI file
In class system_track_type: system_track_number, time_of_information
In class fault_type: event
In class service_health_type: health_state, health_state_reason
In class subsystem_health_type: health_state, health_state_reason
In class downlink_request_type: own_missile_track_id
In class uplink_report_type: own_missile_track_id
In class uplink_request_type: own_missile_track_id

(They are linked to an implicitly created Java datatype by the EA tool)

Fix linkage in the master EA file

For each attribute link the type to the actual class in the model rather than free text.

Text such as
"MethodTag: ea_guid =

"
appears in some tables in the specification.
These should not be there

Remove MethodTag markers in the specification document

Remove all instance of "MethodTag: ea_guid =

" from the tables describing methods

The identifier sensor_plot_equipement_assessment_type has a typo (extra E in equipment) in all places that it is mentioned in the document.

Fix typo in sensor_plot_equipement_assessment_type

Remove the additional 'e' from the name of the type in the model and regenerate all artefacts.

The PSM generation rules lead to a duplicate class name of performance_assessment_request_type in packages for the Sensor Performance use cases (one in domain model and one in the service model). This is undesirable and may lead to confusion and/or erroneous usage of the spec.

Rename performance_assessment_request_type

The class performance_assessment_request_type in the Sensor_Performance (Domain Model) package should be renamed to performance_assessment_parameters_type. This is arguably a better name and disambiguates in the generated PSMs.
Additionally, operations sim_mode_status in Control_Simulation clashes with a Domain Model class in the PSM and operations complete in Perform_Illumination & Perform_Missile_Downlink clash in the generated PSMs. These should be renamed to be more consistent and coherent:

• sim_mode_status to report_sim_mode_status
• complete to report_illumination_completed & report_uplink_completed respectively

The Provide Sensor Orientation use case would be more useful if there was a possibility to provide an origin for the sensor. In particular, it could be used in cases where the sensor is mobile with respect to the physical platform on which it is situated; particular to communicate field of view.
Equally optional coverage in azimuth and elevation would be useful.

Add optional origin and field of view to sensor orientation

To sensor_orientation_type:

• add an optional origin of position_coordinate_type
• add an optional azimuth_coverage of azimuth_interval_type
• add an optional elevation_coverage of elevation_interval_type
• add an optional origin_coordinate_specification of coordinate_specification_type

In the DDS/IDL PSM there is a circular dependency between the Track_Reporting and Sensor_Assessment packages in Sensor_Domain. This is because the by reference association to assessment types contain an assessment_objective_id_type which is defined in the Assessment package.
References in the other direction are intentional.

Move assessment_objective_id_type to Track Reporting

Typedef assessment_objective_id_type should be moved to the Track_Reporting package to resolve the circular dependency.

There is an erroneous attribute "long _id;" in struct concentration_plot_cell_type in Clutter_Reporting.idl

This is a PSM specific issue.

Correct aggregations in the Clutter Reporting domain model

Correct composition of clutter_map_cell_type by clutter_report_type so that it is navigable only from latter to former.

Correct composition of concentration_plot_cell_type by plot_concentration_report_type so that it is navigable only from latter to former.

The diagram files exported from Enterprise Architect for this FTF are in PNG format. These don't scale well as noted in AB review comments.

Export diagrams in SVG format

All the diagrams changed by the FTF should be exported in SVG and saved in a zip file alongside the PNG file that they replaced, so that the OMG editor can use these files to create the specification file for publication.

The enumerator DELETED has been reinstated back in track_phase_type. This does question why the enumerator was renamed NOT_USED in version 1.1, in preference for the use of topic delete. Actually, having reinstated DELETED back in the IDL over using topic delete is a benefit since it makes track lifecycle easier. It is also not clear how INACTIVE assists and how long a track stays in this state before it is LOST. This is not documented but assumed to be sensor specific.

Parameter track_phase_type changed between 1.1 and 2.0. DELETED that was removed in 1.1 is now back!
OARIS 1.1
// The detection lifecycle phase of the track
enum track_phase_type

;

OARIS 2.0
// The detection lifecycle phase of the track
enum track_phase_type

;

Remove DELETED attribute and specific delete operation

The reappearance of the DELETED attribute in the DDS PSM was an error. It was an unintentional consequence of a change of approach to DDS IDL generation.
Having added also added a GraphQL PSM that also has has technology specific convention for deleting data, it makes sense to remove this from the PIM as well. At the PIM level the ability for a sensor to indicate that it has deleted a track is required and should be added. This method would be mapped to DDS API and general GraphQL mapping for the PSMs.
The INACTIVE enumerate was added to the v2.0 proposal for the case of passive sensors where the signal is expected to be intermittent.

I used RTI's DDS tool to transform the OARIS 2.0 beta 1 IDL into C++ and Java and tried to compile it. I got errors involving the enumeration types identity_type in Common_Types.idl and health_state_type in Subsystem_Control.idl. Both have an enumeration literal named UNKNOWN. The generated C++ and Java have these in the same namespace so the code does not compile.

The full error when compiling C++ is below
./Subsystem_Control.h:436:5: error: redeclaration of 'UNKNOWN'
UNKNOWN
^
In file included from ./Subsystem_Control.h:23:0,
from ./Control_Battle_Override.h:24,
from ./Control_Battle_Override.cxx:31:
./Common_Types.h:133:5: note: previous declaration 'org_omg_c4i_Domain_Model_Common_Types_identity_type UNKNOWN'

I'm using rtiddsgen version 5.3.1.14 to transform IDL into C++ and Java. I don't know if this is a problem with the specification or if I need to find a way to transform the IDL into C++ and Java that uses separate namespaces for enumeration types. But I felt someone should know about this anyway.

This was the only issue preventing compilation of the generated C++ and Java. I changed one of the literals to something else and it transformed and built with no errors.

Change health UNKNOWN enum to UNKNOWN_HEALTH

Change to disambiguate the Health enumerate as this does not impact known implementations of the standard. And the values of the Identity enumerate reflect standard usage for identification.

new Optional attributes added to sensor_track_type, to include:
• track_quality
• time_of_first_detection
• time_of_last_detection
• priority
• amplitude
• activity_id
• sensor_function_id
• observed_function_id
• equipment_id
• platform_id
Some appear to be quite low level and applicable to plots rather than tracks. It is also not clear what some of the values are. The last 5 are not documented in the IFS to assist.

Document the mapping of associations in the DDS and GraphQL PSM

All these attributes are optional and whilst not expected to be applicable to (for instance) radar tracks generally, they are applicable (and indeed required) to support the data models of some of the additional sensors scoped by OARIS v2.0 (e.g. EW).
The PSM mapping sections (within section 8) do not discuss the mapping of (by reference) associations. This should be added, which would define the semantics of these attributes in the PSM. (Each of the attributes already has documentation in the PSM in the Beta version).

The parameter subsystem_id has been removed from sensor_track_type. The parameter subsystem_id was added in OARIS 1.1 and provided a few benefits to include ensuring samples being unique since it was used in key. Please can we have it back.

The same issue can be found in sensor_plot_type, perform_cued_search_type and report_cued_search_result_type. Not all interfaces have been checked, but assume the same issue exists elsewhere.

It is assumed that removing subsystem_id is not intentional and in error, since it is documented in the DDS PSM at section 8.2, but does not appear in the PIM in the earlier sections.

Suggest also that subsystem_id_type is changed from unsigned short to unsigned long. This aligns with other UK information models.

Reinstate subsystem_id to PSMs

subsystem_id was omitted from the DDS/IDL PSM mapping in error in the v2.0 proposal (change to PSM generation approach).
This attribute should be added back in to the DDS/IDL PSM files. One keyed attribute per topic type.
Note that the subsystem_id attribute continued to be described as being a key in topic structs in section 8.2.

Request_id_type is defined as unsigned long long 64-bit type, which causes issues with implementation languages and machine base type representation on 64-bit machines.
Uses of request_id_type in several places in the standard are to identify request-response cycles that don't typically operate at high volume or high rate.
So does request_id_type need such a large range? Would unsigned long be sufficient, hence making implementation more machine-friendly?

Change request_id_type to unsigned long

Change the representation of request_id_type to long so that mapping to implementation languages is consistent. (E.g. Java doesn't have unsigned types).
OARIS doesn't have usages of request_id that would exceed a need to generate new/unique values at greater than 1Hz. This supports a system lifetime of nearly 70 years before requiring the negative range.

The subsystem_id has been removed from the perform_cued_search_type and report_cued_search_result_type data types.

For the shape model aspects, a new parameter id of type long has been added to figure_ref_point_type in Shape_Model.idl. This is not documented in the IDL or in the IFS, and the intention is hence not clear.

Remove erroneous _id parameter

A parameter named _id has been generated into the figure_ref_point_type struct. This should be removed. It arises because the aggregation to rectangle_type is navigable both ways. This should be corrected to fix the issue (navigable only from rectangle to figure ref point). This will update diagram 7.29.
There is already an issue to fix/add back subsystem ids generally in the DDS/IDL mapping. This will fix the search topic types as well.

Issues with spec:
OARIS/1.1/PDF, formal/20-08-03

Ambiguous PIM/PSM definition:
Figure 7-41: Track Reporting - Sensor Track (Logical diagram)
Does not include in the sensor_track_type class:
org::omg::c4i::Domain_Model::Common_Types::subsystem_id_type subsystem_id;

Table 7-110: Attributes of IDLStruct sensor_track_type
Does not include :
org::omg::c4i::Domain_Model::Common_Types::subsystem_id_type subsystem_id;

track_phase track_phase_type Track phase (e.g. TRACKED, DELETED, LOST)
DELETED has been removed, but the enumerate renamed NOT_USED

Table 7-111: Attributes of IDLEnum track_phase_type
«idlEnum» DELETED Track has been deleted.
Should be:
«idlEnum» NOT_USED Note: Delete enumeration not used for DDS; dispose topic instance instead.

Duplicate of two separate issues

The issues raised duplicate the contents of two issue raised separately subsequently

The multiplicity for the association between a sensor_track_type class and a sensor_plot_type class is 1..* (one to many) in the model. This doesn't allow for tracks that aren't based on plots (explicitly or implicitly). A zero to many relationship would allow for this and correspondingly offer more flexibility to implementations without unduly compromising rigour.

Change multiplicity for based_on to zero to many

Change the multiplicity for the "based_on" role for association from sensor_track_type to sensor_plot_type to be "0 .. *".
Note that this change doesn't impact the corresponding IDL for the DDS PSM. This is because the lower bound isn't modelled; in both cases it is mapped to a bounded sequence (length 256) according to the role tag in the model (Length = 256).

IDLStruct sensor_track_type does not have notes for all attributes. "* optional *" is applied to only some of the optional attributes.

Add Notes to IDLStruct sensor_track_type

Add notes where missing and remove or replace the "* optional *" markers as these are redundant due to the multiplicity being stated as [0..1]

IDLStruct cued_search_report_type is defined as an optional sensor track id. However sequence diagrams for the Perform Cued Search use case have notes that expect a possibly zero length list of track ids to be returned.

Update diagrams to refer to an optional sensor track id

It is not clear that returning a list of track ids is desirable. In particular it is not clear in general that a sensor would be aware as to when the list was complete.
Therefore the interface method should stay the same but the behavioural model should allow for multiple responses. Subsystems may only implement sending a single response; CMS components may choose to process just the first or each one (and possibly overwrite).

The TrackID parameter of method dedesignate_target has type fire_control_channel_id_type. This is inconsistent. Particularly given it is possible to designate by position rather than by track the name should be FireControlChannelID to match methods receive_target_acquired and receive_fire_control_channel_released.
Actually the name of the receive_target_acquired parameter should be changed from FireControlChannel to FireControlChannelID for consistency too.

Correct Target Designation Parameter Names

In the Process_Target_Designation use case change the parameter names for consistency.

Sequence diagram artefacts such as InteractionOccurrence and ActivityPartition have generated documentation sections as if they were classes in some subsections.
These are supposed to be suppressed by the documentation generation process

Remove sequence diagram artefact subsections

Auto-generated subsections relating to model entity artefacts from sequence diagrams should be removed

There is a leading backslash in the Engagement Support section

Remove typo from Engagement Support heading

remove backslash typo

The process target designation use case in Engagement Support contains a specific error message. This should be deleted as the general use cases method raise_error is used in the interaction diagram. Using the general method is consistent with the design intent and the approach in the other OARIS use cases. (This reflects a failure to apply a design change/improvement consistently by the original submission team).

Remove specific error method

Remove method receive_target_designation_error

The specification shows Register Interest as a prepcondition to doing anything (e.g. figure 7.2). 7.8.5.1 Provide_Sensor_Tracks contradicts this by saying that either Register Interest or Provide Subsystem Services are required. However in at least one real world system using this standard, sensor tracks are successfully received without either of these. The standard should reflect the defacto use and remove registration from cases where it is not requried.

Map Register Interest and Provide Subsystem Services to DDS Discovery

These use cases are implemented by the built-in DDS discovery mechanisms when the DDS PSM is used

Everyting in the IDL has a redundant and innacurate "Version 1.0" attached to it. Many types have documentation that is meaningless, e.g.

            // a simple union type, to represent an optional value
            union sensor_track_position_accuracy_type switch (boolean)

Maintain version number in IDL files and add DDS PSM

Update IDL file template so that the version number reflects the version number of the OARIS standard. (I.e. Version 1.1 as a result of this RTF). This will provide direct trace and is easily achieved as the IDL is generated from the UML model.

The OARIS Specification lacks specific sections for PSM. For other standards there would be a Section 8 describing the DDS PSM and then referring to the IDL files.

"The type of the recognition_type is ‘short.’ This short number is mapped to a recognition_type. " manages to be wrong (it's an unsigned short) while carrying no useful information.

Expand on and correct notes for recognition type

Reflect that the type is unsigned short. Explain that a recognition type is an abstraction for a system or implementation specific taxonomy for the kind of tactical object detected by the track.

A sensor track should carry the identity of the sensor it originates from. Forcing out of band mechanisms (e.g. partition names in the DDS PSM) to be used adds to the complexity of solutions, especially for use cases such as logging where we don't neecssarily care where it came from as long as we know. A compound key would allow all present use cases of filtering etc to be supported.
The situation is particularly confusing for bearing tracks, where the standard apparently requires us to define a bearing with no means of defining the origin.

Add subsystem id attribute to DDS PSM Topic types

Making the source (or destination) of the data explicit in the application payload for DDS makes sense, does not put an unreasonable burden on implementations, can be done with minimal change to existing implementations and is clearly beneficial for the purposes of recording and analysis.
OARIS v1.0 continued to define subsystem_id_type (unsigned short) in package Common_Types, although it is unused by the specification. Its semantics remain applicable for this change proposal.
All DDS Topic types should have a key attribute subsystem_id of subsystem_id_type added. The Subsystem sets it to zero in the receive_sub_identification_data topic and is allocated a value to use in the receive_cms_identification_data response. Note that to disambiguate the multiple subsystems case, the CMS response must be published in a subsystem specific partition.
However, the use of partitions is useful for filtering at a data reader and provides a more robust mechanism for ensuring data is from the subsystem that it labelled as belonging to. Therefore, the partition convention employed in some existing implementations should be formalised within the (new) DDS PSM section.

The name suggests something like degrees latitude per second, degrees longitude per second; probably a usable unit but not one I have ever come across. It was certainly not where I looked when looking for course and speed. I did look at polar_velocity_type, found it was wrong and used cartesian intead, which was not what my subscriber was expecting.

Admittedly it is obvious once one finally drills down to the definiton of wgs84_velocity_type, but experience shows that at least this developer did not search that far.

Add further explanation to velocity_coordinate_type

Explain the nature and purpose of the 3 velocity variants. Add explanation to wgs84_velocity_type that reports velocity from the viewpoint of the object in terms of course and speed with optional angle of climb for changes in height.

It could be speed over ground or speed along direction of travel; with a steep angle of climb the two might be significantly different.

Add notes to the course and speed attributes

Clarify that course is relative to North on the WGS84 spheroid and that speed is the total speed within the spheroid in the direction of travel accounting for angle of climb.

The intended use Cartesian only full covariance classes and accuracy classes for each of the coordinate systems has caused some confusion amongst users of the specification. The intent needs to be more clearly described and explained.

Explain that covariance should be used as an alternative to accuracy values

Some sensors are able to provide a covariance matrix to describe uncertainty in position and velocity calculation as part of their processing. For other sensors accuracy values for each (or perhaps the only) measured quantity are more appropriate.
When constructing a sensor_track_type instance subsystems should use either one strategy or the other. This should be clearly specified in the notes for the covariance matrix, position and velocity accuracy attributes,.

The search use case allows a CMS to instruct a radar subsystem to search a volume of space. This could be used to cue the radar to an existing track. However, it maybe more natural to do this by reference to another sensor track (for instance the radar then has the opportunity to subscribe to updates to the existing track directly to assist in its search).

Add a cue to track method

Add a cue to track method to the search use case that cues by reference to another OARIS sensor track instance rather than to a volume relative to the sensor.

The notes for coordinate_specification_type refer to a NEGOTIATED attribute that does not exist in the relevant enumeration classes.

Remove Reference to NEGOTIATED in Notes

Remove NEGOTIATED text and description from notes. The attribute was deleted by the submitters prior to the final submission, but the notes were not properly cleaned up.

Particularly when reporting bearings from an off-board or otherwise mobile sensor it would be more convenient, natural and precise to report the bearing origin as part of the report rather than referring implicitly to some otherwise well-known origin as the spec currently states (e.g. by reference to another track).

Add Support for Explicit Origin Reference for Bearings

Extend coordinate_origin_type to include an Explicitly Specified enumeration. Add an optional WGS84 origin attribute to polar_position_type and polar_position_accuracy_type. It is only useful to explicitly specify the origin in an absolute coordinate system, if in a relative system the ultimate question of coordinate origin is still unresolved and must rely on implicit information.

When using the DDS PSM the meaning of a topic dispose is sometimes unclear, particularly in the case of a sensor track report as this has a delete/cancel attribute inherited from the PIM.

Remove cancel track attribute from DDS PSM

The DDS PSM in general maps PIM concepts to native DDS mechanisms wherever appropriate. This principle should be applied here and the cancel attribute for IDL struct sensor_track_type omitted from the DDS PSM. However, the enum ordering / positioning should be unaffected so as to maintain compatibility with other potential PSMs.

The classes latitude_coordinate_type and longitude_coordinate_type in Coordinates_and_Positions have a Unit tag = rad (for radians) and a Range from -pi/2 .. pi/2 and -pi .. pi respectively.
The notes for the class state degrees however.

Change Class Tags to be degrees

Change the tags for classes latitude_coordinate_type and longitude_coordinate_type to be the equivalent for degrees as opposed to radians.
This is consistent with implementations and the original submitters intent.

There is an additional_information in the track_reporting use case whilst there is an (abbreviated) additional_info attribute in plot_reporting.

It also appear as additional_information in tracking_control

Change name additional_info attribute name in sensor_plot_type

Change the name of Attribute additional_info in Class sensor_plot_type to additional_information. Update class diagram for sensor plot reporting, the attributes table and the reference in the descriptive text for the class.
This will also result in a corresponding change to the IDL for the DDS PSM.

It has been found that neither of these class names suggest that they contain course and speed as components. Therefore when wishing to report in terms of course and speed a developer failed (initially) to find the correct data structure.

Add WGS84 velocity explanations

This is a duplicate of previously raised issue (same source - different route).

The length of the subsystem parameter string types is contrained to a maximum of 32 characters. This is not long enough to accomodate a hierarchical naming scheme with (potentially long and) meaningful names at each level.
This issue only impacts the IDL for the DDS PSM. The specification at PIM level does not specify a maximum length.

Increase length of parameter_name attributes

Increase the length of the parameter_name attribute from 32 to 128 in the following classes in package Subsystem_Control
name_value_pair_type
parameter_name_type
name_error_pair_type
descriptor

There is a typo (missing 'n') in a parameter for the report_transmission_mode_state operation in the use case Manage Frequency Usage
trasmissionModeSetting should be transmissionModeSetting

Fix typo in parameter name

Change name of parameter to transmissionModeSetting from trasmissionModeSetting (add the missing 'n')

The description of the table for new interface for Sensor Orientation (in proposal 69) has the sensor plot detail in it, instead of the detail for the new interface.

i.e.
Method | Notes | Parameters
write_sensor_plot() | This method receives a individual plot update from the sensor. It is expected to be called periodically from the sensor. | sensor_plot_type plots The set of plots

should be
Method | Notes | Parameters
write_sensor_orientation() | Informs the CMS of the orientation of the sensor. | sensor_orientation_type orientation The orientation of the sensor.

Correct text for sensor orientation

Correct text describing the operations of the interface from that described by proposal 69 (a copy of previous section) to that as intended from the proposal description

Across several of the subsystem control, sensor control and sensor tracking use cases the mechanism for determing subsystem state are inconsistent and some cases do not allow for this without modifying the subsystem's state. CMS functionality will in many cases need to determing current state of the subsystem (without modification) - e.g. to present a control window to an operator.

(extract from Nick Davies email - Thales)
Looking at some of the sensor management and CMS use-cases, there seems to be 3 patterns used to obtain the state of the subsystems amongst the interfaces. Those patterns are essentially:

• Modify/Response
  The CMS issues a modification request to the subsystem. The subsystem will then call the CMS back with the new state (or old if unsuccessful); the state published will only include the attributes that pertain to the original modify request. The callback is mapped to the original request through a request ID attribute that is present in both request and response.
• Request/Response
  The CMS issues a report request to the subsystem. The subsystem will then call the CMS back with the current state; the state published will only include the attributes that pertain to the original report request. The callback is mapped to the original request through a request ID attribute that is present in both request and response.
• Notification
  The subsystem notifies CMS of the current state in an unsolicited fashion, either periodically or on state change

For each of the use-cases in detail:

• ‘Manage Subsystem Parameters’ use-case uses both request/response and modify/response – there is no callback interface for retrieving data.
• ‘Manage Frequency Usage’ use-case uses modify/response and Notification pattern. There is a ‘report_frequencies_state()’ method for notifying frequency state either periodically or on change, and also a ‘transmission_frequency_state’ method that reports the changed frequencies in response to a specific request. For transmission mode, however, there is only the modify/response mechanism – there is no way of discovering the transmission mode without setting a value.
• ‘Control Battle Override’ use-case uses the modify/response mechanism – there is no way of discovering the battle override state without setting a value
• ‘Control Emissions’ use-case uses the modify/response mechanism – there is no way of discovering the emissions state without setting a value first
• ‘Manage Technical State’ use-case uses all three patterns – there is a modify/response interface, a request/response interface and a pure notify interface
• ‘Manage Operational Mode’ use-case uses the modify/response mechanism and the request/response mechanism. However, the interface specifies that the response interface can also be used for ‘spontaneous’ mode change, where a request_id of 0 will be specified.
• ‘Manage Tracking Zones’ use-case uses the modify/response mechanism
• ‘Manage Transmission Sectors’ use-case uses the modify/response mechanism

As you can see, only ‘Manage Frequency Usage’ (part) and ‘Manage Technical State’ provide an interface that allow the subsystems to report their state unsolicited by the CMS. ‘Manage Operational Mode’ provides a definition of where it can provide state information unsolicited, but the standard only states that it will be used when the subsystem needs to change its own state; it does not say that it will also be used when it receives a request to change state from the CMS.

For the other use-cases, there are the following options:

• For the use-cases where the state is atomic, assume that the last value on the topic is the current value, not matter what request caused it to be populated. This can only be used for ‘Control Battle Override’ , ‘Control Emissions’, ‘Manage Frequency Usage’ (other part), ‘Manage Operational Mode’ – the others can allow only part of the state to be reported back, depending on what request initiated the published value. This still relies on the subsystem publishing its initial value on start-up.
• We can assume that all the interfaces use the approach specified in the ‘Manage Operational Mode’ use-case where the full state is reported on a request_id of 0. But then why not just define the interface without the request_id, always reporting all values?
• Where a request/response interface is specified, the CMS can poll periodically for the state. This only helps us with ‘Manage Subsystem Parameters’ out of those not covered in the options so far.

That leaves ‘Manage tracking Zones’ and ‘Manage Transmission Sectors’ where none of our options give us a mechanism for discovering the underlying state without changing it.

Emphasise CMS responsibility to initiate determination of initial state

Add paragraph below to the relevant use case descriptions (as identified in the issue).

"It is the CMS's responsibility to initiate the determination of initial state by making a request for information to the subsystem."

Following the revision to the DDS PSM mappings Manage Mastership now has empty topic types - this is illegal IDL. See OARIS-5 & OARIS-17.

Add count to manage mastership method

Add parameter count : unsigned long to methods acquire_mastership and release_mastership.

The revised DDS mappings lead to empty topic types in the Heartbeat Signal use case - this is illegal IDL. See OARIS-5 & OARIS-17

Add count to receive heartbeat method

Add parameter count : unsigned long to methods receive_subsystem_heartbeat_signal and receive_cms_heartbeat_signal

The system track class is now being used as a topic type and thus requires a key to differentiate individual instances. (See OARIS-3 & OARIS-13).

Add key stereotype to attribute system_track_number

The attribuite system_track_number for class system_track_type is to have a stereotype of "key". Consequently this attribute will be a key for the topic in the DDS PSM due to the PIM to PSM transformation mapping.

The receive_error method from the common_use_case interface is not used by some of the use case service interfaces, but rather these interfaces redefine such a method themselves. The service interface use cases are:
Provide_Health_State
Perform_Missile_Downlink
Perform_Illumination
Perform_Missile_Uplink
Register_Interest

Update services to use common use case method

Service Interfaces for use cases Provide_Health_State, Perform_Missile_Downlink, Perform_Illumination, Perform_Missile_Uplink and Register_Interest are to be updated to remove methods that are duplicates of the common use case interface.

The stand alone C++ mapping for general_polar_volume_type does not compile (at least for v6.4 of OpenSplice). This is because the fields within the union have the same name as as their type (also declared in that namespace). Therefore access functions hide the struct declaration below the point at which they are declared.

Suffix all Shape_Model class names with _type to make unique

Suffix each class name in package Shape_Model with '_type'. This is in accordance with the REG's modelling guidelines and is consistent with the rest of the specification. (Note that general_polar_volume_type already has _type suffix - no change required in that respect).

There is an operation in the CMS interface for use case Manage Frequency Usage named transmission_frequency_state and a type within the Sensor Control domain model named transmission_frequency_state_type.
In the DDS PSM, with the generated mappings, this leads to two types named transmission_frequency_state_type. This is at best confusing and can lead to code compilation difficulties (they are generated into different, but related, namespaces).

Rename manage frequency usage method as a response

Rename method transmission_frequency_state transmission_frequency_state_response to remove the ambiguity.

Track Extractor functions may need to know where the (rotating) sensor which is feeding the measurements is pointing at any moment in time. E.g. to know when plots on a particular bearing would be expected to be received. There is not an interface in the OARIS specification to do this, but it is functionally coupled to the provision of plots so the plot_reporting summary use case is a natural home for it.

Create Provide Sensor Orientation use case

A new use case within the Plot Reporting summary use case is created (alongside Provide Plots) to provide the sensor orientation to functions such as track extraction. The use case is placed at compliance level 2.
Following the modeling framework for OARIS, the use case interface (for the CMS) has a single method to provide the orientation of the sensor. The sensor orientation describes its angle (azimuth and optional elevation), the coordinate orientation (using enumerate from Common Types) and its time of validity.
The coordinate orientation allows for absolute or relative interpretation of angles and defines the zero point for elevation.
(XMI for the updated model is attached together with a document containing views of updated and new diagrams).

The UML model for OARIS (from which the specification is generated) contains two state transition diagrams to show the mandatory and superset of transitions between technical states. However, the specification does not include these diagrams (because of the generation template used). The specification consequently lacks some of the intended semantic detail for the class technical_state_type leading to a lack of understanding from users.

Add state transition diagrams to technical state type

The document generation should be amended to include state transition diagrams that are members of a class.
The missing state transition diagram will thus be included in the specification.

The naming convention for OARIS is to suffix all classes with "_type". This was laid down in the REG's modelling guidelines. The following classes don't follow this guideline: -
frequencies_set_request
freqency_range_set
all_frequencies_state
reported_frequency_state
selected_frequency_list
transmission_frequency_state

Rename Sensor Control Classes to have "_type" suffix

All classes in the Sensor Control domain model should follow the convention and have the _type suffix. Consistency will aid understanding and usability (particularly at the programming level).
This impacts: -
frequencies_set_request (to be deleted by proposal 58)
freqency_range_set (to be deleted by proposal 58)
all_frequencies_state
reported_frequency_state
selected_frequency_list
transmission_frequency_state
and also from other use cases: -
test_target_scenario_state
control_emission_state
test_target_scenario_common_parameter_target
test_target_scenario_independent_target
test_target_plus_scenario
test_target
test_target_scenario_id
test_target_scenario

In the Manage Frequency Usage use case, if the CMS requests a range of frequencies rather than a discrete set (see class frequency_set_request) then the response from the sensor subsystem does not properly model the subsystem adopting the requested state. Class selected_frequency_list contains a list of discrete frequencies. In particular a list with two frequencies in it is indistiguishable from the expected response to a request to use just those two discrete frequencies - as opposed to the range inbetween them.

Frequency band to designate a discrete or range of frequencies

The specification already contains an abstraction of the actual frequencies to use through the frequency band typedef. This class allows a discrete value to be mapped to particular frequencies supported by the sensor. The interface would be simplified without loss of flexibility or rigour by developing the semantics of the frequency_band_type such that it may refer to a discrete frequency or a frequency range.
The frequency_range_set class is then redundant (as a band may refer to a range) - a range of bands does not in general make sense as it would pre-suppose some logical organisation of the discrete frequency band type that it need not have. Consequently, the frequency_set_request union class is also redundant as there would only be the single way of representing frequency state on the interface as opposed to the current two styles; removing the union is a useful simplification.
The set_frequencies() method should both take a selected_frequency_list parameter mirroring the transmission_frequency_state() method.
Metadata pertaining to a frequency band (e.g. whether refers to a discrete frequency or a range) can be communicated using the Manage Subsystem Parameterts use case.

It is not clear from the design of the Manage Frequency Usage use case how the set of frequencies supported by the sensor is determined by the CMS.

Describe usage of report_frequencies_state to determine supported set

Update use case descriptive text and figure 7.104 to describe CMS potential usage of report_frequencies_state to determine the supported set (whether available, enabled or otherwise) before further modification is made.

The attribute sector_added :Boolean in class transmission_sector_type has no descriptive text. The semantics of this attribute are not readily deduced, nor can a coherent proposal for what they should be for all instances of the class be formed. For instance, it should be noted that the class is used both for the CMS to request a setting and for the Sensor to report its current state. Equally, there are not attributes to indicate that a sector has been or should be removed.

Remove attribute sector_added

Attribute sector_added from class transmission_sector_type should be removed as it is both ill-defined and also unnecessary. Both the CMS and a Sensor are able to deduce which sectors have been added, updated or removed (if necessary for their processing) as each sector has its own unique sector_id. Also, having an attribute sector_added etc. makes it possible for the data to be inconsistent - not a good feature.

The description of the Manage Transmission Sectors use case states that sectors with a particular transmission mode can be defined. However, this is not supported in the PIM model - for instance no suitable attribute on transmission_sector_type. Also there is a typo on the bullet heading for Transmission Mode Sectors within the description for this use case.

Add transmission mode to transmission sector type

By adding transmission mode to transmission sector type the transmission mode of a sector can be defined. The value of AUTOMATIC is suitable as a default value for the case where the CMS does not need to constrain the sensor.

These are generic fields used to describe responses following requests. They are published asynchronously to the original request – the request and response are associated with each other using a common request ID found in each.
The response information is a free-text String. Though this can represent useful information to convey directly to an operator, it does not allow a specific field to be identified at fault (the original request may be composed of a number of fields). Also, there may be several fields at fault.
I would suggest the following –
o Allow multiple denial/error reason fields to be associated with the response rather than just one
o Add an identifier to the field, which can be a generic string. The identity will be appropriate to the original response e.g. the parameter name at fault with appropriate namespacing, or a null value if the error message refers to the whole request in general (e.g. loss of communication). The format of the error identifier should be defined in the description of the request service/topic.
(From Nick Davies - Thales UK)

Allow multiple error reasons with parameter identifier

A rejecting request_ack should be able to refer to the particular fields that are at fault and to be able to identify multiple discrete errors (for requests with sets of parameters in particular - e.g. Manage Subsystem Parameters and Manage Transmission Sectors).
A new struct class denial_type with attributes for the existing reason and zero-to-many parameter reference (new typedef of string) should be added to the model. request_ack_type then composes zero-to-many of these.
<<idlStruct>> denial_type
+ reason : denial_reason_type
+ related_parameter : parameter_reference_type [0..*]

<<idlTypedef>> parameter_reference_type
(generalizes string)
Tag: Length = 64

compostion of denial_type by request_ack_type
denial_type role: name = element, multiplicity 0..*

The sensor_track_type class does not allow a track to not have a defined environment value. Whilst the environment may be known or assumed for most cases of a naval radar track, for other types of sensor this is not necessarily the case.

Environment should be optional in sensor_track_type

Make attribute environment optional in sensor_track_type - environment_type then continues to enumerate actual environments and so it is possible to use this type in a non-optional (mandatory) context to define an object as being in a definite environment.

It is inconsistent to define an identity_type field for use with system_track_type class but to then not have the option to use it with the sensor_track_type class.

Add an identity attribute to sensor_track_type

Add an optional sensor_track_pre_identification attribute with type identity_type to sensor_track_type. This is the sensor's view of the tracked object's identity for cases where a sensor has such a capability.

The operation of the Manage Subsystem Parameters use case requires that the names of the parameters are unique within the scope of a particular subsystem. This is not well described in the beta specification.

Add uniqueness notes to description of parameter_name values

The Manage Subsystem Parameter names use case requires parameter names to be unique within the scope of a subsystem.
This should be documented for each applicable attribute in the spec (domain model classes) and in the description of the use case (service model interface)

The IDL for this struct has an '_id' member. This is due to generation of the IDL from the PIM, with the PIM having set the navigability for the aggregation within service_indication_list_type in the wrong direction.

Correct navigability of service indication set composition

Set navigability to be from service_indication_set_type to service_indication_type (and not in reverse direction).

(This will prevent _id attribute being generated in updated set of IDL for DDS PSM).

Class descriptor, supporting use case Manage Subsystem Parameters, contains an attribute called range. This is a reserved word in Ada and leads to compilation errors in code generated from the OARIS model.

Change name of attribute to parameter_range

Change name of attribute range in class descriptor in package Domain_Model.Subsystem_Domain.Subsystem_Control to parameter_range.

Changing the name to parameter_range is both more descriptive and consistent with other attribute names for the type (parameter_name, parameter_type & parameter_unit). It also avoids clashes with names from other language specifications.

identification_response_type from the domain model, supporting the Provide Subsystem Idenitification use case, contains an 'accept' enumeration. This is a reserved word in Ada and hence automatically generated code does not compile.

Fix by using Common Use Case Interface

Remove problem classes and use Common Use Case Interface methods instead. (As per REG's Modelling Guidelines).

There are no #pragma keylist statements in the OARIS IDL. Whilst this statement does not appear to be part of the formal DDS specification, the specification recognises that there does need to be a way of doing this. By way of example the ALMAS PSM does contain keylist statements.
However, X-Types introduces @Key as a syntax for indicating keys - better to do this.

Define key fields to define object uniqueness semantics

Attributes and roles with a <<Key>> stereotype should be defined in the Domain Model so as to define object unqiueness where such concepts exist for the defined classes.
Classes with such semantics are sensor_track_type and possibly sensor_plot_type. Objects of sensor_plot_type are never updated, therefore all occurrence can be treated as separate instances (objects). Therefore a key is necessary for implementation and is not to be applied to sensor_plot_type (this is convenient as the sensor_plot_id_type is defined as optional).
For topic types in general the IDL clearly identifies the Topic types and their mapping to PIM operations in the preceding comments. Therefore no further action is required for the identification and definition of OARIS topic types in general.

Heartbeat Signal in PIM but Exchange Heartbeat in Conformance and in the name of a sequence diagram

Change Conformance section to align with PIM

Change Exchange Heartbeat to Heartbeat Signal in Conformance section and update diagram name as identified by the issue. Do this by updating the EA model

This use case is referenced in the activity diagrams and the compliance levels but is (no longer) present in the PIM services

Remove all references to the Control Interface Connection use case

Delete references to the use case on pages identified by Issue 1 by updating the EA model.
Where used as a condtion for the entry state of an activity diagram, it should be replaced with the Provide Subsystem Services use case (consistent with the Compliance Level 3B diagram for instance).

The PSM maping for sensor_track_set_type is a list of ids, whereas the operation of the interface requires that it is a list of sensor track values.

Change relation from aggregation to composition

A composition is a by-value relation and so the PSM mapping generates a by-value sequence for a composition

There was modelling rule (from the REG's proposal for OARIS) that relationship multiplicities targeting sequences in the IDL PSM all had upper bounds defined. This was achieved by using a Length Tag. There is no such Tag for the 'based on' association role relating sensor_track_type to sensor_plot_type. As a consequence, implementations cannot bound the number of plots that a track may be based on.

Implementation decision to constrain number of plots a track can be based on

It shall be an implementation decision to constrain the number of plots a track can be based on. Including such a constraint in the standard may reduce the standard's applicability and doesn't overly reduce interoperability.

The semantics of Sensor Track Set in the Service Model are that it should contain each Sensor Track by value. However, the DDS / IDL PSM models the class as a sequence of int - i.e. by reference to sensor track instances. Reporting by set is an alternative to reporting individual instances so it is not possible to refer to separately published instances by reference to their id/key value. (Nor does it make any sense to do so).
This is because the relatiohship in the model is aggregation rather than compoistion and the transformation used maps this to a 'by reference' mechanism rather than a 'by value'.

Change sensor track set relation to composition

Change the sensor track set types aggregation of class sensor track type to be a composition (by value).

This changes the generated IDL for the DDS PSM to generate a by value sequence, which is the intended semantic

There was modelling rule (from the REG's proposal for OARIS) that relationship multiplicities targeting sequences in the IDL PSM all had upper bounds defined. This was achieved by using a Length Tag. There is no such Tag for the 'element' association role relating sensor_track_set type to sensor_track_type. As a consequence, implementations cannot bound the number of tracks that may be contained in a track set.

Implementation decision to constrain size of track sets

It shall be an implementation decision to constrain the size of track sets. Including such a constraint in the standard may reduce the standard's applicability and doesn't overly reduce interoperability.

In package Subsystem Control, the class Service Indication Type has an unintended '_id' attribute generated. This is because the composition on the list type is navigable in the reverse direction - leading to the generation of an unintended reference to the list, when the PIM is transformed. The composition between service list type and service name type has a similar problem in the PIM, but being an enumeration the transformation does not generate an extra attribute.

Reverse navigability of service indication set type composition

The navigability of the composition of service indication type by class service indication set type should be from service indication set type to service indication type.

Fixing this will prevent the generation or erroneous _id field in the DDS IDL PSM

Class descriptor in package Subsystem Control has an association with itself with a 0..* related parameter role. This is mapped to int[] in the IDL for the DDS PSM (and array of integers). The meaning of these integers is not documented in the PSM.

Related Parameter to mean index of parameter in descriptor sequence

The related_parameter relation is well defined by UML in the PIM. The related_parameter_id attribute in the DDS PSM IDL should have a comment that this refers to the index of the related_parameter struct within the same descriptor_sequence.element sequence.

Interfaces Provide_Subsystem_Identification_CMS and Provide_Subsystem_Identification_Sub both contain a receive_cms/sub_response method instead of using the common_use_case_interface.

Use common use case interface receive acknowledgement in Provide Subsystem Identification

Remove the use case specific classes and methods and use the common use case interface instead (as per the REG's modelling guidelines). This also means that language name clashes are avoided as well.
The sequence diagrams for the use case's service interface are updated.

Interface Manage_Subsystem_Parameters_CMS contains method receive_error. The method in the common_use_case_interface should be used

Use common use case interface receive error in Manage Subsystem Parameters

The operation receive_error should be removed from interface Manage_Subsystem_Parameters_CMS. In the use cases sequence diagrams the receive error messages should be mapped to the operation from the common_use_case_interface (This has the same signature, so won't be apperent on diagrams only through property dialogs in the model).

An 'EARTH_REFERENCED' value is required for the WGS84 and ECEF coordinate systems.

Add EARTH_REFERENCED enumeration literal to coordinate_origin_type

Add EARTH_REFERENCED enumeration literal to enumeration coordinate_origin_type. This value signifies that the origin for the coordinate system is well-defined with respect to the Earth by the coordinate system. E.g. centre of the Earth for Earth-Centred-Earth-Fixed or the WGS84 spheroid for WGS84.

NORTH_EAST_DOWN enumerate in coordinate_orientation_type is described as EAST_NORTH_DOWN . Also does not state applicability to Cartesian Coordinate System and EAST_NORTH_DOWN does not either (and should)

Update documentation of coordinate_orientation_type enumerates

Describe NORTH_EAST_DOWN as x = north, y = east, z = down
State that NORTH_EAST_DOWN and EAST_NORTH_DOWN apply to Cartesian coordinate systems

Subsystem Ids are introduced in the DDS IDL PSM (by the tool PIM to PSM transformation). They are added to each topic type (generated from each interface method).
DDS Partitions are a more flexible mechanism for logically partitioning multiple publishers and subscribers for the same topic.
Note that this is an issue raised on the DDS IDL PSM, it does not affect the PIM. The PIM considers it to be a PSM concern as to how multiple actors for the subsystem role in particular are distiguished.

Remove subsystem id insertion from PSM generation

The DDS PSM generation inserts a subsystem_id attribute into each topic type. This functionality should be removed, resulting in IDL without subsystem_id attributes, simplifying the translation of the PIM to DDS. Note that this does not impact the PIM.

Key fields need to be attributes of topic types (at least for some DDS implementations). Sensor Track Id is a member of Sensor_Track_Id_Type, which is not the top-level topic type.

Issue fixed by proposal for Issue 3

Fixed by proposal 13.
Using the //@Key notation from X-Types means that lower level types can have keys.

The message connectors in the Heartbeat Signal Basic Flow sequence diagram are modelled as Signals rather than Calls and are not mapped to the interface operations as is the standard for the other use cases

Fix Heartbeat Signal Sequence Diagrams

Update the sequence diagrams for use case Heartbeat Signal to use calls not signals and map the messages to interface operations.