Preprin t 0 (1998) 1{28 1 Spine Routing in Ad Ho c Net w - PDF document

Preprin t 0 (1998) 1{28 1 Spine Routing in Ad Ho c Net w orks Ragh upath y Siv akumar Bev an Das V aduvur Bhargha v an Co or dinate d Scienc es L ab or atory, University of Il linois at Urb ana-Champ aign E-mail: f siv akumr,das,bhargha v g @timely .crhc.uiuc.edu An ad ho c net w ork is a m ulti-hop wireless net w ork in whic h mobile hosts comm u- nicate without the supp ort of a wired bac kb one for routing messages. W e in tro duce a self or ganizing net w ork structure called a spine and prop ose a spine-b ase d r outing infr astructur e for routing in ad ho c net w orks. W e prop ose t w o spine routing algorithms : (a) Optimal Spine R outing (OSR) , that uses full and up-to-date kno wledge of the net w ork top ology , and (b) Partial- know le dge Spine R outing (PSR) , that uses partial kno wledge of the net w ork top ol- ogy . W e analyze the t w o algorithms and iden tify the optimalit y-o v erhead trade-o�s in v olv ed in these algorithms. Keyw ords: Ad Ho c Net w orks, Spine Routing, Wireless Net w orks 1. In tro duction An ad ho c net w ork is a m ultihop wireless net w ork in whic h mobile hosts comm unicate o v er a shared, scarce wireless c hannel. Ad ho c net w orks lac k a wired bac kb one to main tain routes as hosts mo v e, or turn o� or on. Instead, the hosts in the ad ho c net w ork co op erativ ely determine routes in a distributed manner. Th us, ev ery host is p oten tially a router, and a route b et w een t w o hosts ma y c hange not only b ecause of end-host mobilit y but also b ecause of in terme- diate router mobilit y . The k ey issues that an y routing algorithm for suc h an en vironmen t m ust address are the dynamics of the top ology and the o v erhead of route computation. Among the w ell kno wn paradigms for routing, shortest p ath algorithms and algorithms are at the t w o extremes of the r outing ondemand sp ectrum. T ypically , shortest path routing pro vides optimal routes but incurs a h uge o v erhead in the pro cess (in the form of p erio dic or ev en t-driv en up dates of link c hanges o v er the en tire net w ork). Routing on-demand, on the other hand, t ypically has lo w o v erhead but computes p ossibly non-optimal or stale routes, and requires �o o ding the net w ork when no route is curren tly a v ailable.

2 / Giv en that most of the recen t w ork on routing in ad ho c net w orks can b e classi�ed in to one of the ab o v e t w o mec hanisms, w e p erceiv e a need for a more e�cien t algorithm that w ould balance the optimalit y of the shortest path algo- rithms with the lo w o v erhead of the routing on-demand algorithms. T o w ards this end, w e in tro duce a self or ganizing net w ork structure called a spine that helps com bine optimalit y with lo w o v erhead. Brie�y , the spine is c hosen to b e a small and relativ ely stable subnet w ork of the ad ho c net w ork, and is used to aggregate the net w ork state distributiv ely among the no des of the spine in order to generate routes. Using this spine structure, w e presen t a new spineb ase d r outing infr astruc tur e for fault-toleran t unicast and m ulticast routing in ad ho c net w orks. The k ey issues addressed in this pap er are: (a) ho w to build and main tain the spine, (b) what net w ork top ology information to collect in the spine, and (c) ho w to compute routes once the information is aggregated in the spine no des. This in- frastructure is sp eci�cally built to address the dynamics of the net w ork top ology , scarcit y of the shared resources, and the nature of applications that ma y t ypically run in suc h adho c net w orking en vironmen ts. W e address the follo wing goals in this pap er: � Supp ort e�cien t unicast routing b y trading-o� b et w een shortest path routing on-demand algorithms. � Supp ort m ulticast routing b y using the spine structure as the m ulticast bac k- b one. � Compute alternate routes for long-liv ed connections, and switc h routes dy- namically up on failure of the primary route in order to pro vide fault-toleran t routing. Note that the spine is basically a routing infrastructure. Its primary role is not to carry data tra�c. Ho w ev er, the spine can b e used to transmit data pac k ets temp orarily when routes are b eing switc hed or computed, or for m ulticast �o ws. F or the spine-based routing infrastructure describ ed ab o v e, w e presen t t w o spine routing algorithms: (a) Optimal Spine R outing OSR , that uses full and up-to-date kno wledge of the net w ork top ology , and (b) Partialknow le dge Spine R outing PSR , that uses partial kno wledge of the net w ork top ology . F or practical

/ 3 and scalable ad ho c net w orks, w e prop ose a com bination of and a clustered PSR algorithm called CSR in [3]. Through CSR , w e pro vide a hierarc hical structure for the spine in order to scale w ell for large net w orks. Since ad ho c net w orks are exp ected to b e naturally clustered [10], a hierarc hical organization is b oth useful and e�cien t. The rest of the pap er is organized as follo ws. Section 2 presen ts the net w ork mo del and notation for our w ork. Section 3 describ es the Optimal Spine Routing algorithm. Section 4 describ es the P artial-kno wledge Spine Routing algorithm. Section 5 ev aluates the p erformance of OSR and PSR. Section 6 discusses some issues and applications of spine routing. Section 7 concludes the pap er. 2. Net w ork Mo del Communic ation mo del W e assume that mobile hosts in an ad ho c net w ork use omnidirectional transmitters on a common wireless c hannel. All hosts within transmission range R of the transmitting host ma y receiv e the transmission (in absence of hidden stations). Hence, eac h transmission can either b e an unreliable lo c al br o adc ast in tended for all receiv ers within range, or a reliable unic ast in tended for a single receiv er. Gr aph terminolo gy and notation W e use an undirected graph G = ( V ; E ), with m edges and n no des, to represen t a snapshot of the ad ho c net w ork. Eac h no de in V represen ts a mobile host, and eac h edge in E signi�es that t w o hosts are within transmission range of eac h other. The top olo gy of G is the set of edges and no des. Hence, when w e sa y a c hanges the top ology , w e mean a c hange in the net w ork that no de movement results in a c hange in either V or E . Sp eci�cally , a e dge deletion o ccurs when t w o hosts lose comm unication with eac h other, and an e dge insertion o ccurs when t w o hosts mo v e in to range of eac h other. A in isolation o ccurs when no de deletion a host turns o� its p o w er, and a no de insertion in isolation o ccurs when a host turns on its p o w er. By \in isolation" w e mean that no other c hange has o ccurred in the net w ork. Because a no de insertion or deletion a�ects m ultiple edges, w e pro cess these c hanges to V as m ultiple c hanges to E . Finally , the most general no de movement mo dels the mo v emen t of a host from one part of the net w ork to