You could configure static routes in s1, s2, s3 and s4 for 126.96.36.199/24 and 188.8.131.52/24 with the dedicated links as next-hops. You should not configure ospf in those links, and the static routes should be configured with no-readvertise. Also make sure that in case of a link failure the interfaces will go down, if not configure bfd for static routes. The subnets 184.108.40.206/24 and 220.127.116.11/24 should be redistributed into OSPF, so this way in case the dedicated links go down, you will still have reachability through the other links.
The reason for this proposal is because you cannot modify ospf routing behaviour inside an area. So there is no way to have ospf in that link and make it only advertise/accept a certain IP subnet. The protocol static preference is lower than ospf, hence the static route will be preferred, and in case the link fails, static route will be removed from the RIB and the ospf route will be used.
As mentioned above, OSPF will not allow the behavior you want. So you will need to change routing protocols for the links.
Besides the static routes mentioned above you could switch to BGP. You would use local preference and your export policies to insure that the link you want to use as the primary path for prefixes is the preferred one on each side.
Steve Puluka BSEET - Juniper Ambassador IP Architect - DQE Communications Pittsburgh, PA (Metro Ethernet & ISP) http://puluka.com/home
One way to achieve this requirement is to run dedicated OSPF process over red links. JUNOS calls it "routing-instance type no-forwarding". If 3.3.3/24 and 4.4.4/24 are in default OSPF process, redistribute only those prefixes into that dedicated process.