Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
The Event Language
Search
Sponsored
·
Your Podcast. Everywhere. Effortlessly.
Share. Educate. Inspire. Entertain. You do you. We'll handle the rest.
→
Andreas Pfohl
June 09, 2015
Research
0
570
The Event Language
Rule Based Complex-Event Processing on Distributed Networks
Andreas Pfohl
June 09, 2015
Tweet
Share
More Decks by Andreas Pfohl
See All by Andreas Pfohl
DNSSEC v2
apfohl
0
55
Kore
apfohl
0
230
DNSSEC
apfohl
0
170
Domain Name System
apfohl
1
230
FreeBSD
apfohl
0
260
Other Decks in Research
See All in Research
生成AI による論文執筆サポート・ワークショップ 論文執筆・推敲編 / Generative AI-Assisted Paper Writing Support Workshop: Drafting and Revision Edition
ks91
PRO
0
120
社内データ分析AIエージェントを できるだけ使いやすくする工夫
fufufukakaka
1
880
超高速データサイエンス
matsui_528
2
380
HU Berlin: Industrial-Strength Natural Language Processing with spaCy and Prodigy
inesmontani
PRO
0
200
地域丸ごとデイサービス「Go トレ」の紹介
smartfukushilab1
0
900
ForestCast: Forecasting Deforestation Risk at Scale with Deep Learning
satai
3
380
Akamaiのキャッシュ効率を支えるAdaptSizeについての論文を読んでみた
bootjp
1
440
20年前に50代だった人たちの今
hysmrk
0
140
SREのためのテレメトリー技術の探究 / Telemetry for SRE
yuukit
13
3k
病院向け生成AIプロダクト開発の実践と課題
hagino3000
0
530
世界モデルにおける分布外データ対応の方法論
koukyo1994
4
870
Combining Deep Learning and Street View Imagery to Map Smallholder Crop Types
satai
3
560
Featured
See All Featured
CoffeeScript is Beautiful & I Never Want to Write Plain JavaScript Again
sstephenson
162
16k
Mozcon NYC 2025: Stop Losing SEO Traffic
samtorres
0
140
Noah Learner - AI + Me: how we built a GSC Bulk Export data pipeline
techseoconnect
PRO
0
100
Bootstrapping a Software Product
garrettdimon
PRO
307
120k
16th Malabo Montpellier Forum Presentation
akademiya2063
PRO
0
49
Why Our Code Smells
bkeepers
PRO
340
58k
Leveraging LLMs for student feedback in introductory data science courses - posit::conf(2025)
minecr
0
140
Build The Right Thing And Hit Your Dates
maggiecrowley
38
3k
We Analyzed 250 Million AI Search Results: Here's What I Found
joshbly
1
690
We Are The Robots
honzajavorek
0
160
Mobile First: as difficult as doing things right
swwweet
225
10k
The browser strikes back
jonoalderson
0
360
Transcript
The Event Language Rule Based Complex-Event Processing on Distributed Networks
Andreas Pfohl and Robert Heumüller June, 9th 2015 1
Setting • Complex-Event Processing • Embedded Networks • Different Architectures
• Varying Computational Power 2
Challenges • High-level language • Runs on different architectures •
Compiled and Just-in-Time-Compiled • Means of distribution 3
Event Processing event PositionEvent { time = [0], pos =
[1] } pos_a; event PositionEvent { time = [2], pos = [6] } pos_b; + event VelocityEvent { v = [2.5] } velocity; 4
Vector Calculation • Vectors for information • Vector Calculator •
Collections Library 5 event PositionEvent { time = [0], pos = [1] } pos_a; f = 1.123; v = [1, 2, f]; s = 3 * v; u = v + s;
Event Language 6 Event Definition: event TimedEvent { time };
event PositionEvent extends TimedEvent { position }; event VelocityEvent { velocity };
Event Language 7 Function Definition: VelocityEvent calculateVelocity(PositionEvent posEv1, PositionEvent posEv2)
:= { velocity = (posEv2.time - posEv1.time)^(-1) * (posEv2.position - posEv1.position) };
Event Language 8 Predicate Definition: predicate positionCheck(PositionEvent posEv1, PositionEvent posEv2)
:= posEv2.position > posEv1.position; predicate timeCheck(PositionEvent posEv1, PositionEvent posEv2) := posEv2.time > posEv1.time;
Event Language 9 Rule Declaration: VelocityRule: [PositionEvent, PositionEvent : positionCheck,
timeCheck] -> calculateVelocity;
Event Language 10 event TimedEvent { time }; event PositionEvent
extends TimedEvent { position }; event VelocityEvent { velocity }; VelocityEvent calculateVelocity(PositionEvent posEv1, PositionEvent posEv2) := { velocity = (posEv2.time - posEv1.time)^(-1) * (posEv2.position - posEv1.position) }; predicate positionCheck(PositionEvent posEv1, PositionEvent posEv2) := posEv2.position > posEv1.position; predicate timeCheck(PositionEvent posEv1, PositionEvent posEv2) := posEv2.time > posEv1.time; VelocityRule: [PositionEvent, PositionEvent : positionCheck, timeCheck] -> calculateVelocity;
Challenges • High-level language • Runs on different architectures •
Compiled and Just-in-Time-Compiled • Means of distribution 11
LLVM • Formerly Low Level Virtual Machine • Compiler Infrastructure
• Compiled and JIT • Different Target Architectures 12
Challenges • High-level language • Runs on different architectures •
Compiled and Just-in-Time-Compiled • Means of distribution 13
Challenges • High-level language • Runs on different architectures •
Compiled and Just-in-Time-Compiled • Means of distribution 14
Challenges • High-level language • Runs on different architectures •
Compiled and Just-in-Time-Compiled • Means of distribution 15
Implementation 16
Overview 17
Lexer / Parser Toolchains: • Lex / Yacc • Flex
/ Bison • Antlr • … 18 Jobs: • Parse Language • Build AST Flex / Lemon
Lemon • Context Free • LALR(1) Parser • Robust Syntax
• Error Handling • LibCollect AST 19 rule_declaration(NODE) ::= TYPE(T) COLON rule_signature(RS) RARROW IDENTIFIER(I). { struct payload *payload = malloc(sizeof(struct payload)); payload->type = N_RULE_DECLARATION; payload->alternative = ALT_RULE_SIGNATURE; payload->rule_declaration.name = strdup(T); payload->rule_declaration.identifier = strdup(I); NODE = tree_create_node(payload, 1, RS); }
Scoping 20 VelocityEvent calculateVelocity(PositionEvent posEv1, …) := …
Linking 21 Resolve References
Validation • Parser: Syntactic Validity • Validation: Semantic Validity •
Check: • References Resolved • Type Constraints 22 Unexpected Behavior Invalid Input
Type Checking 23 Assert: Function_Definition.Expression == EVENT Assert: Initializer.Value ==
NUMBER
Code Generation Tasks: • For each event: • Structure definition
• For each Predicate / Function: • Function • For each Rule: • Activation Function • Processing Function 24 Options: • C • “Highlevel” • Complex • No JIT • Assembler • Lowlevel • Less Complex • Plattform Specific LLVM is Middle Ground
LLVM • Intermediate Representation • Provides Optimizer • C-API 25
Meets all Requirements
LLVM-IR 26 %PositionEvent = type <{ i16, double*, i16, double*
}> %VelocityEvent = type <{ i16, double* }> define %VelocityEvent* @VelocityRule_function(%PositionEvent*, %PositionEvent*) { %3 = call %VelocityEvent* @calculateVelocity(%PositionEvent* %0, %PositionEvent* %1) ret %VelocityEvent* %3 } define %VelocityEvent* @calculateVelocity(%PositionEvent*, %PositionEvent*) { %3 = alloca %VelocityEvent* %malloccall = tail call i8* @malloc(i32 ptrtoint (%VelocityEvent* getelementptr (%VelocityEvent* null, i32 1) to i32)) %4 = bitcast i8* %malloccall to %VelocityEvent* %5 = getelementptr inbounds %VelocityEvent* %4, i32 0, i32 0 %6 = alloca double %7 = alloca <{ i16, double* }> %8 = getelementptr inbounds <{ i16, double*
Conclusion • Rule-Based language for Event-Processing • Created complete compiler
toolchain • Knowledge about compiler construction • Improved low level programming and understanding 27
28 Demo - Creating Event-Processor
Demo - Using Event-Processor