"... In PAGE 33: ... The following summary lists suggested directions and motivations for several broad levels at which synthesis for MEMS should be developed. To organize the findings and recommendations, Table1 shows the different levels of MEMS design. The synthesis process involves traversing this figure from the top system level down to the bottom process- mask level.... In PAGE 105: ... Hence, there is a great need to develop computer aided analysis tools and system level simulators that are easily usable by the MEMS community[4, 5]. Discussion Table1 contrasts different aspects of VLSI systems, macro (traditional) mechanical sys- tems, and MEMS. Structured design methods exist for VLSI systems, because they operate in a single... In PAGE 106: ... not as complex at present. Kinematics plays a big Predominantly monolithic role compliant structures Manufactured with Wide range of Same as VLSI systems planar lithography manufacturing techniques including 3-D machining Table1... ..."

"... In PAGE 4: ....1.Single Mix Protection Profile The Single Mix PP was written to address the security problems of a single mix system, without considering the requirements of the user (who wants to send anonymous mail) and also ignoring all the security threats that may derive from the connection of the system with other mixes. Thus the threat list of this PP ( Table3 ) includes items like flooding attacks, logical access to the TOE, re- play attacks, traffic size analysis. The last two threats (marked with TE ) are intended to be countered not by the mix itself but by the environment (operating system, etc.... ..."

"... In PAGE 11: ...Table2 . The numbers in the 1st column refer to the local fault effects in Table 1.... In PAGE 11: ... It can be the starting point of criticality analysis. The 2nd column in Table2 shows a possible grouping of global effects into criticality categories. A trip based on bad data is more critical (3) than not creating the trip at all (factor 2).... In PAGE 11: ... It is intended as a solution to improve the dependability of the system. The suggested protection mechanisms are described in the 4th column of Table2 . The interpretation of the 3rd row is the following: if the tour operator does not enter the username and password, the process will wait and hang infinitely.... In PAGE 12: ...the fault tolerance measures listed in Table2 . In the modified process all activ- ities that have an input could also finish the process because of timeout.... ..."

"... In PAGE 18: ...11 The gap analysis is a key step in a larger programme. While there is no space to discuss the whole Pro- gramme of Work here, a summary of the main outputs is given in Table3 below, with steps most directly related to the gap analysis highlighted in Figure 5. This is one of the most ambitious environmental programmes ever attempted by the international community and is uniting government agencies with NGOs and many other stakeholders in efforts to meet the far-reaching goals and tight deadlines.... ..."

"... In PAGE 6: ... The Soft-Error Protection: Test Results side- bar discusses radiation testing of some soft-error protection techniques. Table2 shows a compara- tive analysis of these techniques with respect to several system-level metrics, exploring some vari- ables and factors that determine their applicabil- ity to actual designs. REUSE PARADIGM FOR BUILT-IN SOFT-ERROR RESILIENCE A new paradigm that leverages the reuse of on- chip resources for multiple functions at various stages of manufacturing and field use can over- come the drawbacks of existing soft-error protec- tion techniques.... ..."

"... In PAGE 17: ... 2 4.2 Schematic Type Synthesis An algorithm for schematic type synthesis is given by the rules of Table6 . It is a system for deriving judgements of the form ? ` M ) X; C.... In PAGE 17: ... Intuitively, X; C schematically represent the set of types for M in ?. The algorithm makes use of two auxiliary functions, CUM and *, de ned in Table6 . These functions are analogous to the functions cum and quot; of Table 4, and are characterized by the following lemmas.... In PAGE 19: ...ypej+i. Let be extended with 7! j + i. If X 6wh Typej for any j, then X 6wh Type for any , so take = . 2 The rules of Table6 make use of an informal convention whereby level variables are required to be \new quot;. This means that the level variable cho- sen at that rule occurrence is unique to that occurrence, and di erent from that associated with any occurrence of any other rule in the derivation un- der consideration.... In PAGE 20: ...whether or not there exists a schematic term X and consistent con- straint set C such that ? ` M ) X; C is derivable. Proof (Soundness) The rst property is proved by inspection of the rules of Table6 . For the second property, consider a derivation of ? ` M ) X; C.... In PAGE 20: ... More precisely, we build a derivation of ? ` M ) X by induction on the height of . The induction proceeds by case analysis of the root node of based on the rules of Table6 . The most interesting case is when the root of is an instance of rule (a-app).... In PAGE 21: ...4 applies to extend A;B to the required cum([N=x]A2;i). (Decidability) The proof is by induction on the structure of M, keeping in mind that the rules of Table6 are syntax-directed. The base cases (Prop, Typei, and variables) are all trivial: for the case of a variable x, we need... In PAGE 35: ... ` Prop ) Type ; f 0 g ( new ) a-d-type ` Typei ) Type ; f gt; i g ( new ) a-d-var x[x:A] x ` x ) CUM x(A; ;) a-d-def x[x=M:X; G] x ` x ) LV(G)(X; G) a-d-gen ` A ) X; C ` X wh 1 x 62 Dom( ) C consistent [x:A] ` B ) Y; D ` Y wh 2 ` fx:AgB ) 1 *C[D 2 a-d-abs ` A ) X; C ` X wh x 62 Dom( ) C consistent [x:A] ` M ) Y; D ` [x:A]M ) fx:AgY; C [ D a-d-app ` M ) X; C ` X wh fx : X1gX2 ` N ) Y; D ` X1 # Y (E) ` MN ) CUM ([N=x]X2; C [ D [ E) where V assigns \new quot; level variables to each of the level variables in V, * is as in Table6 , and CUM is de ned by CUM (X; C) = CUM( (X); C). Table 10: Type Synthesis Algorithm for CC! with De nitions... In PAGE 39: ... ; C ` Prop ) Prop; Type ; f 0 g ( new ) a-ad-type ; C ` Typej ) Typej; Type ; f gt; j g ( new ) a-ad-anon ; C ` Type ) Type ; Type ; f gt; 0 g ( ; new ) a-ad-var x[x:X] x; C ` x ) x; CUM(X; ;) a-ad-def x[x=X:Y; G] x; C ` x ) LV(G)nLV(C)(X; Y; G) a-ad-gen ; C ` Q ) U; X; D X wh 1 C [ D consistent [x:U]; C [ D ` R ) V; Y; E Y wh 2 x 62 Dom( ) ; C ` fx:QgR ) fx:UgV; 1 *D[E 2 a-ad-abs ; C ` Q ) U; X; D X wh C [ D consistent [x:U]; C [ D ` R ) V; Y; E x 62 Dom( ) ; C ` [x:Q]R ) [x:U]V; fx:UgY; D [ E a-ad-app ; C ` Q ) U; X; D X wh fx : X1gX2 ; C ` R ) V; Y; E X1 # Y (F) ; C ` QR ) UV; CUM([V=x]X2; D [ E [ F) where * and CUM are as de ned in Table6 , and V is as de ned in Table 10. Table 12: Algorithm for Anonymous Universes and De nitions... ..."