1: \begin{abstract}

2:   The basic adjoint relationship (BAR) approach is an analysis technique based on the stationary equation of a Markov process.   This approach was introduced to study heavy-traffic, steady-state

3:   convergence of generalized Jackson networks in which each service station has a

4:   single job class. We extend it to multiclass queueing networks operating under

5:   static-buffer-priority (SBP) service disciplines.    Our extension makes a connection with Palm

6:   distributions that allows one to attack a difficulty arising from

7:   queue-length truncation, which appears to be unavoidable  in the multiclass

8:   setting.

9: 

10:   For multiclass queueing networks operating under

11:   SBP service disciplines, our BAR approach 

12:   provides an alternative to the ``interchange of limits'' approach

13:   that has dominated the literature in the last twenty years.

14:   The BAR approach can produce sharp results and

15:   allows  one to establish  steady-state convergence under three additional

16:   conditions: stability, state space collapse (SSC) and a certain matrix

17:   being ``tight.''  These three conditions do not appear to depend on

18:   the interarrival and service-time distributions beyond their means,

19:   and their verification can be studied as three separate modules. In

20:   particular, they can be studied in a simpler, continuous-time Markov

21:   chain setting when all distributions are exponential.

22: 

23:   As an example,

24:   these three conditions are shown to hold in reentrant lines operating

25:   under last-buffer-first-serve discipline.  In a two-station,

26:   five-class reentrant line, under the heavy-traffic condition, the

27:   tight-matrix condition implies both the stability condition and the SSC

28:   condition.  Whether such a relationship holds   generally is an open problem.

29: \end{abstract}

30: