Here's a solution that doesn't depend on SQL Server 2012:

declare @test-2 table

(Tsq INT IDENTITY (1,1),

Data Varchar (150),

ts datetime,

Tpkt_type int)

insert into @test-2 values ('GS,000020,000021,000022,000023','2013-11-13 09:47:35.963','2') --1

insert into @test-2 values ('NS,000020,000021,000022,000023','2013-11-13 09:47:36.180','2')

insert into @test-2 values ('GS,000020,000021,000022,000023','2013-11-13 09:47:37.007','2')

insert into @test-2 values ('GX,1,0000000000000000000000000','2013-11-13 09:47:37.007','4') --1

insert into @test-2 values ('GS,000020,000021,000022,000023','2013-11-13 09:50:25.987','2') --2

insert into @test-2 values ('GS,000020,000021,000022,000023','2013-11-13 09:50:26.190','2')

insert into @test-2 values ('GS,000020,000021,000022,000023','2013-11-13 09:50:26.393','2')

insert into @test-2 values ('GX,1,0000000000000000000000000','2013-11-13 09:50:40.920','4') --2

insert into @test-2 values ('GS,000020,000021,000022,000023','2013-11-13 09:51:28.330','2') --3

insert into @test-2 values ('GS,000020,000021,000022,000023','2013-11-13 09:51:28.547','2')

insert into @test-2 values ('GS,000020,000021,000022,000023','2013-11-13 09:51:28.767','2')

insert into @test-2 values ('GX,1,0000000000000000000000000','2013-11-13 09:51:43.257','4') --3

insert into @test-2 values ('NS,000020,000021,000022,000023','2013-12-17 16:51:09.063','18') --4

insert into @test-2 values ('GS,000020,000021,000022,000023','2013-12-17 16:51:09.063','2')

insert into @test-2 values ('GS,000020,000021,000022,000023','2013-12-17 16:51:09.063','2')

insert into @test-2 values ('GX,1,0000000000000000000000000','2013-12-17 16:51:15.257','4') --4

;with

GX as (

select *, row_number() over(order by ts) rn

from @test-2

where Data like 'GX%'

),

buckets as (

Select gx1.ts gxbegin, gx2.ts gxend, gx2.data gxdata, gx2.Tpkt_type gxtype

from GX gx1

right join gx gx2

on gx2.rn = gx1.rn+1

),

bucketed as (

select t.*, b.* from @test-2 t

join buckets b

on t.ts < b.gxend and (t.ts > b.gxbegin or b.gxbegin is null)

),

filtered as (

select t.*, b.gxdata, b.gxend, b.gxtype from buckets b

cross apply (

select top(1) * from @test-2 t

where t.ts < b.gxend

and (t.ts > b.gxbegin or b.gxbegin is null)

order by t.ts asc

) t

)

select * from filtered