Identified bugs in the current (official) version of SEAMCAT-2

 

The following provides a list of all bugs identified in the latest version of SEAMCAT-2. Since after switch to SEAMCAT-3 platform the ERO will no longer provide maintenance and revisions of SEAMCAT-2 source code, these bugs will not be removed from the SEAMCAT-2 software. Therefore all users are invited to keep these bugs in mind when using SEAMCAT-2. On the other hand, ERO checked that these bugs were not transfered to SEAMCAT-3 during the porting of original source code.

Please also note that some of the earlier identified deficiencies of the current SEAMCAT software version 2.1.0 are reflected in the SEAMCAT User Manual, available for downloads elsewhere on this site.

***

20 September 2005 - Complete-1 algorithm in Interference Calculation Engine

Another remaining bug was identified in the SEAMCAT-2 software, notably in its Interference Calculation Engine. When the Complete-1 algorithm is used in SEAMCAT-2 in cases where the generated useful dRSS signal lies closely to the Victim Receiver sensitivity level, then the Complete-1 simulation results may produce erroneous results due to the fact that the snapshots with dRSS<VRsensitivity are not discounted as they should have been.

This may be easily checked by setting the VR sensitivity level above the dRSS range, this should normally mean that the wanted link is not functional in itself, hence interference probability should be 0 regardless of iRSS values, however the simulations with Complete-1 algorithm in SEAMCAT-2 will still produce some probability result, depending only on distance between dRSS vs iRSS vectors.

On the other hand, if dRSS vector range in your simulated scenarios falls well above the VR sensitivity level, then Complete-1 calculation results are correct. Also the Quick and Complete-2 algorithms in SEAMCAT-2 are unaffected of this bug.

***

8 January 2004 - Discrete uniform distribution

It was noted that the use of "Discrete uniform" distribution function may generate incomplete sets of modelled parameter values when setting its Step to less than 1. In particular, the discrete uniform function does not assign parameter values corresponding to i=N for some N values (N is the number of steps, defined as N=(Xmax-Xmin)/Step, ref. Annex 1 of User Manual).

As an illustration of this, example in section 2.3.2 of User Manual may be taken, where Victim link frequency is set as Discrete uniform function with Xmin=449.9875 MHz; Xmax=450.1125 MHz and Step=0.025. This set-up should produce N=5 steps, where the Victim system is in turn assigned frequencies [450.0; 450.025; 450.05; 450.075; 450.1] MHz. However detailed analysis shows that in this case the distribution function never assigns last step frequency 450.1 MHz to the Victim link. Similar erroneous behaviour reappears when Xmin and Xmax are re-set so as to give N=3, but distribution works properly for other N values, e.g. when N=2, 4, 6.

This erroneous performance also happens with other Step<1 values, although not necessarily for the same N. E.g. when using discrete distribution with Step=0.1, the function would work erroneously for N=2 or 5, while N=3, 4, 6 would not produce errors.

Given the ongoing re-design of the SEAMCAT software, correction of this bug is left for the new version of the software, due for release summer 2004. Meanwhile all SEAMCAT users are urged to double check their cases of using the Discrete uniform function. The appearance of this bug may be avoided by using other types of distribution producing identical results, e.g. "User-defined (Stair)" function. In this case, the above described scenario with particular frequency hopping set-up may be replaced by defining manually the discrete (Stair) distribution with N=5 as shown in the attached picture.

This also means that the results of calculations (Figs. 12 and 13) in the Example 2 given in section 2.3.2 of User Manual were actually not correct. This is corrected in the new version of the User Manual, dated 8 January 2004 and available for download on the main SEAMCAT page.

***

15 July 2003 - Intermodulation interference

When calculating impact of intermodulation, the SEAMCAT produces a vector of Intermodulation Interfering Received Signal Strength (iRSSintermod), with instant values expressed in relation to the Victim Receiver's noise floor (see Annex 9 of the User Manual).

However, when the Interference Calculation is later initiated, its software engine assumes that the iRSSintermod values are expressed in absolute (dBm) terms, as it is the case for two other interfering signal vectors (unwanted and blocking). Therefore, the calculation of probability of intermodulation for selected signal type "Intermodulation" will produce erroneous results.

This fault is noted for correction in the next release of SEAMCAT software. Meanwhile, it appears that users may apply a temporary "work-around" technique - that is to increase the value of intermodulation attenuation parameter of Victim receiver by one third of the receiver's noise floor value. This puts the resulting calculated iRSSintermod values at about their absolute values, therefore the interference calculation engine may derive the correct interference probability.

***

 

For any queries related to information on this page or to SEAMCAT in general please contact:

Jean-Philippe Kermoal (ERO) via e-mail to "kermoal---ero.dk" or phone +45 33 89 63 07

 

 
 
 
 

Go to ECO main page Print current page Send e-mail to ECO Map of this site Browse Meeting Documents Info on use of web

 
 
Related Links  

  WG SE main page

  STG main page

  SEAMCAT Forum

  SEAMCAT-3:
  
- Free Download
   - Online Manual (Wiki)

   - Tutorials/Materials

   - Source code 

   - FAQ 

   - Practical tips 

  E-mail reflector:
  For SEAMCAT users:
  SEAMCAT E-mail reflector

  For SEAMCAT contributors:
  STG E-mail reflector